mesa/src/mapi/glapi/gen/glX_proto_send.py

1123 lines
32 KiB
Python
Raw Normal View History

Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
#!/usr/bin/env python
# (C) Copyright IBM Corporation 2004, 2005
# All Rights Reserved.
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# on the rights to use, copy, modify, merge, publish, distribute, sub
# license, and/or sell copies of the Software, and to permit persons to whom
# the Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice (including the next
# paragraph) shall be included in all copies or substantial portions of the
# Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
# IBM AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
# IN THE SOFTWARE.
#
# Authors:
# Ian Romanick <idr@us.ibm.com>
# Jeremy Kolb <jkolb@brandeis.edu>
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
import gl_XML, glX_XML, glX_proto_common, license
import sys, getopt, copy, string
def convertStringForXCB(str):
tmp = ""
special = [ "ARB" ]
i = 0
while i < len(str):
if str[i:i+3] in special:
tmp = '%s_%s' % (tmp, string.lower(str[i:i+3]))
i = i + 2;
elif str[i].isupper():
tmp = '%s_%s' % (tmp, string.lower(str[i]))
else:
tmp = '%s%s' % (tmp, str[i])
i += 1
return tmp
def hash_pixel_function(func):
"""Generate a 'unique' key for a pixel function. The key is based on
the parameters written in the command packet. This includes any
padding that might be added for the original function and the 'NULL
image' flag."""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
h = ""
hash_pre = ""
hash_suf = ""
for param in func.parameterIterateGlxSend():
if param.is_image():
[dim, junk, junk, junk, junk] = param.get_dimensions()
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
d = (dim + 1) & ~1
hash_pre = "%uD%uD_" % (d - 1, d)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if param.img_null_flag:
hash_suf = "_NF"
h += "%u" % (param.size())
if func.pad_after(param):
h += "4"
n = func.name.replace("%uD" % (dim), "")
n = "__glx_%s_%uD%uD" % (n, d - 1, d)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
h = hash_pre + h + hash_suf
return [h, n]
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
class glx_pixel_function_stub(glX_XML.glx_function):
"""Dummy class used to generate pixel "utility" functions that are
shared by multiple dimension image functions. For example, these
objects are used to generate shared functions used to send GLX
protocol for TexImage1D and TexImage2D, TexSubImage1D and
TexSubImage2D, etc."""
def __init__(self, func, name):
# The parameters to the utility function are the same as the
# parameters to the real function except for the added "pad"
# parameters.
self.name = name
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.images = []
self.parameters = []
self.parameters_by_name = {}
for _p in func.parameterIterator():
p = copy.copy(_p)
self.parameters.append(p)
self.parameters_by_name[ p.name ] = p
if p.is_image():
self.images.append(p)
p.height = "height"
if p.img_yoff == None:
p.img_yoff = "yoffset"
if p.depth:
if p.extent == None:
p.extent = "extent"
if p.img_woff == None:
p.img_woff = "woffset"
pad_name = func.pad_after(p)
if pad_name:
pad = copy.copy(p)
pad.name = pad_name
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.parameters.append(pad)
self.parameters_by_name[ pad.name ] = pad
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.return_type = func.return_type
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.glx_rop = ~0
self.glx_sop = 0
self.glx_vendorpriv = 0
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.glx_doubles_in_order = func.glx_doubles_in_order
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.vectorequiv = None
self.output = None
self.can_be_large = func.can_be_large
self.reply_always_array = func.reply_always_array
self.dimensions_in_reply = func.dimensions_in_reply
self.img_reset = None
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.server_handcode = 0
self.client_handcode = 0
self.ignore = 0
self.count_parameter_list = func.count_parameter_list
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.counter_list = func.counter_list
self.offsets_calculated = 0
return
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
class PrintGlxProtoStubs(glX_proto_common.glx_print_proto):
def __init__(self):
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
glX_proto_common.glx_print_proto.__init__(self)
self.name = "glX_proto_send.py (from Mesa)"
self.license = license.bsd_license_template % ( "(C) Copyright IBM Corporation 2004, 2005", "IBM")
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.last_category = ""
self.generic_sizes = [3, 4, 6, 8, 12, 16, 24, 32]
self.pixel_stubs = {}
self.debug = 0
return
def printRealHeader(self):
print ''
print '#include <GL/gl.h>'
print '#include "indirect.h"'
print '#include "glxclient.h"'
print '#include "indirect_size.h"'
2006-10-13 14:28:48 +01:00
print '#include "glapi.h"'
print '#include "glthread.h"'
print '#include <GL/glxproto.h>'
2005-05-07 18:14:50 +01:00
print '#ifdef USE_XCB'
2006-10-09 16:25:03 +01:00
print '#include <X11/Xlib-xcb.h>'
print '#include <xcb/xcb.h>'
print '#include <xcb/glx.h>'
2005-05-07 18:14:50 +01:00
print '#endif /* USE_XCB */'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print ''
print '#define __GLX_PAD(n) (((n) + 3) & ~3)'
print ''
self.printFastcall()
self.printNoinline()
print ''
print '#ifndef __GNUC__'
print '# define __builtin_expect(x, y) x'
print '#endif'
print ''
print '/* If the size and opcode values are known at compile-time, this will, on'
print ' * x86 at least, emit them with a single instruction.'
print ' */'
print '#define emit_header(dest, op, size) \\'
print ' do { union { short s[2]; int i; } temp; \\'
print ' temp.s[0] = (size); temp.s[1] = (op); \\'
print ' *((int *)(dest)) = temp.i; } while(0)'
print ''
print """NOINLINE CARD32
__glXReadReply( Display *dpy, size_t size, void * dest, GLboolean reply_is_always_array )
{
xGLXSingleReply reply;
(void) _XReply(dpy, (xReply *) & reply, 0, False);
if (size != 0) {
if ((reply.length > 0) || reply_is_always_array) {
const GLint bytes = (reply_is_always_array)
? (4 * reply.length) : (reply.size * size);
const GLint extra = 4 - (bytes & 3);
_XRead(dpy, dest, bytes);
if ( extra < 4 ) {
_XEatData(dpy, extra);
}
}
else {
(void) memcpy( dest, &(reply.pad3), size);
}
}
return reply.retval;
}
NOINLINE void
__glXReadPixelReply( Display *dpy, struct glx_context * gc, unsigned max_dim,
GLint width, GLint height, GLint depth, GLenum format, GLenum type,
void * dest, GLboolean dimensions_in_reply )
{
xGLXSingleReply reply;
GLint size;
(void) _XReply(dpy, (xReply *) & reply, 0, False);
if ( dimensions_in_reply ) {
width = reply.pad3;
height = reply.pad4;
depth = reply.pad5;
if ((height == 0) || (max_dim < 2)) { height = 1; }
if ((depth == 0) || (max_dim < 3)) { depth = 1; }
}
size = reply.length * 4;
if (size != 0) {
void * buf = malloc( size );
if ( buf == NULL ) {
_XEatData(dpy, size);
__glXSetError(gc, GL_OUT_OF_MEMORY);
}
else {
const GLint extra = 4 - (size & 3);
_XRead(dpy, buf, size);
if ( extra < 4 ) {
_XEatData(dpy, extra);
}
__glEmptyImage(gc, 3, width, height, depth, format, type,
buf, dest);
free(buf);
}
}
}
#define X_GLXSingle 0
NOINLINE FASTCALL GLubyte *
__glXSetupSingleRequest( struct glx_context * gc, GLint sop, GLint cmdlen )
{
xGLXSingleReq * req;
Display * const dpy = gc->currentDpy;
(void) __glXFlushRenderBuffer(gc, gc->pc);
LockDisplay(dpy);
GetReqExtra(GLXSingle, cmdlen, req);
req->reqType = gc->majorOpcode;
req->contextTag = gc->currentContextTag;
req->glxCode = sop;
return (GLubyte *)(req) + sz_xGLXSingleReq;
}
NOINLINE FASTCALL GLubyte *
__glXSetupVendorRequest( struct glx_context * gc, GLint code, GLint vop, GLint cmdlen )
{
xGLXVendorPrivateReq * req;
Display * const dpy = gc->currentDpy;
(void) __glXFlushRenderBuffer(gc, gc->pc);
LockDisplay(dpy);
GetReqExtra(GLXVendorPrivate, cmdlen, req);
req->reqType = gc->majorOpcode;
req->glxCode = code;
req->vendorCode = vop;
req->contextTag = gc->currentContextTag;
return (GLubyte *)(req) + sz_xGLXVendorPrivateReq;
}
const GLuint __glXDefaultPixelStore[9] = { 0, 0, 0, 0, 0, 0, 0, 0, 1 };
#define zero (__glXDefaultPixelStore+0)
#define one (__glXDefaultPixelStore+8)
#define default_pixel_store_1D (__glXDefaultPixelStore+4)
#define default_pixel_store_1D_size 20
#define default_pixel_store_2D (__glXDefaultPixelStore+4)
#define default_pixel_store_2D_size 20
#define default_pixel_store_3D (__glXDefaultPixelStore+0)
#define default_pixel_store_3D_size 36
#define default_pixel_store_4D (__glXDefaultPixelStore+0)
#define default_pixel_store_4D_size 36
"""
for size in self.generic_sizes:
self.print_generic_function(size)
return
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
def printBody(self, api):
self.pixel_stubs = {}
generated_stubs = []
for func in api.functionIterateGlx():
if func.client_handcode: continue
# If the function is a pixel function with a certain
# GLX protocol signature, create a fake stub function
# for it. For example, create a single stub function
# that is used to implement both glTexImage1D and
# glTexImage2D.
if func.glx_rop != 0:
do_it = 0
for image in func.get_images():
if image.img_pad_dimensions:
do_it = 1
break
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if do_it:
[h, n] = hash_pixel_function(func)
self.pixel_stubs[ func.name ] = n
if h not in generated_stubs:
generated_stubs.append(h)
fake_func = glx_pixel_function_stub( func, n )
self.printFunction(fake_func, fake_func.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.printFunction(func, func.name)
if func.glx_sop and func.glx_vendorpriv:
self.printFunction(func, func.glx_vendorpriv_names[0])
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.printGetProcAddress(api)
return
def printGetProcAddress(self, api):
procs = {}
for func in api.functionIterateGlx():
for n in func.entry_points:
if func.has_different_protocol(n):
procs[n] = func.static_glx_name(n)
print """
#ifdef GLX_SHARED_GLAPI
static const struct proc_pair
{
const char *name;
_glapi_proc proc;
} proc_pairs[%d] = {""" % len(procs)
names = procs.keys()
names.sort()
for i in xrange(len(names)):
comma = ',' if i < len(names) - 1 else ''
print ' { "%s", (_glapi_proc) gl%s }%s' % (names[i], procs[names[i]], comma)
print """};
static int
__indirect_get_proc_compare(const void *key, const void *memb)
{
const struct proc_pair *pair = (const struct proc_pair *) memb;
return strcmp((const char *) key, pair->name);
}
_glapi_proc
__indirect_get_proc_address(const char *name)
{
const struct proc_pair *pair;
/* skip "gl" */
name += 2;
pair = (const struct proc_pair *) bsearch((const void *) name,
(const void *) proc_pairs, ARRAY_SIZE(proc_pairs), sizeof(proc_pairs[0]),
__indirect_get_proc_compare);
return (pair) ? pair->proc : NULL;
}
#endif /* GLX_SHARED_GLAPI */
"""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def printFunction(self, func, name):
footer = '}\n'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if func.glx_rop == ~0:
print 'static %s' % (func.return_type)
print '%s( unsigned opcode, unsigned dim, %s )' % (func.name, func.get_parameter_string())
print '{'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
else:
if func.has_different_protocol(name):
if func.return_type == "void":
ret_string = ''
else:
ret_string = "return "
func_name = func.static_glx_name(name)
print '#define %s %d' % (func.opcode_vendor_name(name), func.glx_vendorpriv)
print '%s gl%s(%s)' % (func.return_type, func_name, func.get_parameter_string())
print '{'
print ' struct glx_context * const gc = __glXGetCurrentContext();'
print ''
print '#if defined(GLX_DIRECT_RENDERING) && !defined(GLX_USE_APPLEGL)'
print ' if (gc->isDirect) {'
print ' const _glapi_proc *const table = GET_DISPATCH();'
print ' PFNGL%sPROC p =' % (name.upper())
print ' (PFNGL%sPROC) table[%d];' % (name.upper(), func.offset)
print ' %sp(%s);' % (ret_string, func.get_called_parameter_string())
print ' } else'
print '#endif'
print ' {'
footer = '}\n}\n'
else:
print '#define %s %d' % (func.opcode_name(), func.opcode_value())
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print '%s __indirect_gl%s(%s)' % (func.return_type, name, func.get_parameter_string())
print '{'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if func.glx_rop != 0 or func.vectorequiv != None:
if len(func.images):
self.printPixelFunction(func)
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.printRenderFunction(func)
elif func.glx_sop != 0 or func.glx_vendorpriv != 0:
self.printSingleFunction(func, name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
pass
else:
print "/* Missing GLX protocol for %s. */" % (name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print footer
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def print_generic_function(self, n):
size = (n + 3) & ~3
print """static FASTCALL NOINLINE void
generic_%u_byte( GLint rop, const void * ptr )
{
struct glx_context * const gc = __glXGetCurrentContext();
const GLuint cmdlen = %u;
emit_header(gc->pc, rop, cmdlen);
(void) memcpy((void *)(gc->pc + 4), ptr, %u);
gc->pc += cmdlen;
if (__builtin_expect(gc->pc > gc->limit, 0)) { (void) __glXFlushRenderBuffer(gc, gc->pc); }
}
""" % (n, size + 4, size)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def common_emit_one_arg(self, p, pc, adjust, extra_offset):
if p.is_array():
src_ptr = p.name
else:
src_ptr = "&" + p.name
if p.is_padding:
print '(void) memset((void *)(%s + %u), 0, %s);' \
% (pc, p.offset + adjust, p.size_string() )
elif not extra_offset:
print '(void) memcpy((void *)(%s + %u), (void *)(%s), %s);' \
% (pc, p.offset + adjust, src_ptr, p.size_string() )
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
else:
print '(void) memcpy((void *)(%s + %u + %s), (void *)(%s), %s);' \
% (pc, p.offset + adjust, extra_offset, src_ptr, p.size_string() )
def common_emit_args(self, f, pc, adjust, skip_vla):
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
extra_offset = None
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
for p in f.parameterIterateGlxSend( not skip_vla ):
if p.name != f.img_reset:
self.common_emit_one_arg(p, pc, adjust, extra_offset)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if p.is_variable_length():
temp = p.size_string()
if extra_offset:
extra_offset += " + %s" % (temp)
else:
extra_offset = temp
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def pixel_emit_args(self, f, pc, large):
"""Emit the arguments for a pixel function. This differs from
common_emit_args in that pixel functions may require padding
be inserted (i.e., for the missing width field for
TexImage1D), and they may also require a 'NULL image' flag
be inserted before the image data."""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if large:
adjust = 8
else:
adjust = 4
for param in f.parameterIterateGlxSend():
if not param.is_image():
self.common_emit_one_arg(param, pc, adjust, None)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.pad_after(param):
print '(void) memcpy((void *)(%s + %u), zero, 4);' % (pc, (param.offset + param.size()) + adjust)
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
[dim, width, height, depth, extent] = param.get_dimensions()
if f.glx_rop == ~0:
dim_str = "dim"
else:
dim_str = str(dim)
if param.is_padding:
print '(void) memset((void *)(%s + %u), 0, %s);' \
% (pc, (param.offset - 4) + adjust, param.size_string() )
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if param.img_null_flag:
if large:
print '(void) memcpy((void *)(%s + %u), zero, 4);' % (pc, (param.offset - 4) + adjust)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
else:
print '(void) memcpy((void *)(%s + %u), (void *)((%s == NULL) ? one : zero), 4);' % (pc, (param.offset - 4) + adjust, param.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
pixHeaderPtr = "%s + %u" % (pc, adjust)
pcPtr = "%s + %u" % (pc, param.offset + adjust)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if not large:
if param.img_send_null:
condition = '(compsize > 0) && (%s != NULL)' % (param.name)
else:
condition = 'compsize > 0'
print 'if (%s) {' % (condition)
print ' (*gc->fillImage)(gc, %s, %s, %s, %s, %s, %s, %s, %s, %s);' % (dim_str, width, height, depth, param.img_format, param.img_type, param.name, pcPtr, pixHeaderPtr)
print '} else {'
print ' (void) memcpy( %s, default_pixel_store_%uD, default_pixel_store_%uD_size );' % (pixHeaderPtr, dim, dim)
print '}'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
else:
print '__glXSendLargeImage(gc, compsize, %s, %s, %s, %s, %s, %s, %s, %s, %s);' % (dim_str, width, height, depth, param.img_format, param.img_type, param.name, pcPtr, pixHeaderPtr)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def large_emit_begin(self, f, op_name = None):
if not op_name:
op_name = f.opcode_real_name()
print 'const GLint op = %s;' % (op_name)
print 'const GLuint cmdlenLarge = cmdlen + 4;'
print 'GLubyte * const pc = __glXFlushRenderBuffer(gc, gc->pc);'
print '(void) memcpy((void *)(pc + 0), (void *)(&cmdlenLarge), 4);'
print '(void) memcpy((void *)(pc + 4), (void *)(&op), 4);'
return
def common_func_print_just_start(self, f, name):
print ' struct glx_context * const gc = __glXGetCurrentContext();'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
# The only reason that single and vendor private commands need
# a variable called 'dpy' is becuase they use the SyncHandle
# macro. For whatever brain-dead reason, that macro is hard-
# coded to use a variable called 'dpy' instead of taking a
# parameter.
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
# FIXME Simplify the logic related to skip_condition and
# FIXME condition_list in this function. Basically, remove
# FIXME skip_condition, and just append the "dpy != NULL" type
# FIXME condition to condition_list from the start. The only
# FIXME reason it's done in this confusing way now is to
# FIXME minimize the diffs in the generated code.
if not f.glx_rop:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
for p in f.parameterIterateOutputs():
if p.is_image() and (p.img_format != "GL_COLOR_INDEX" or p.img_type != "GL_BITMAP"):
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print ' const __GLXattribute * const state = gc->client_state_private;'
break
print ' Display * const dpy = gc->currentDpy;'
skip_condition = "dpy != NULL"
elif f.can_be_large:
skip_condition = "gc->currentDpy != NULL"
else:
skip_condition = None
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.return_type != 'void':
print ' %s retval = (%s) 0;' % (f.return_type, f.return_type)
if name != None and name not in f.glx_vendorpriv_names:
print '#ifndef USE_XCB'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.emit_packet_size_calculation(f, 0)
if name != None and name not in f.glx_vendorpriv_names:
print '#endif'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
condition_list = []
for p in f.parameterIterateCounters():
condition_list.append( "%s >= 0" % (p.name) )
# 'counter' parameters cannot be negative
print " if (%s < 0) {" % p.name
print " __glXSetError(gc, GL_INVALID_VALUE);"
if f.return_type != 'void':
print " return 0;"
else:
print " return;"
print " }"
if skip_condition:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
condition_list.append( skip_condition )
if len( condition_list ) > 0:
if len( condition_list ) > 1:
skip_condition = "(%s)" % (string.join( condition_list, ") && (" ))
else:
skip_condition = "%s" % (condition_list.pop(0))
print ' if (__builtin_expect(%s, 1)) {' % (skip_condition)
return 1
else:
return 0
def printSingleFunction(self, f, name):
self.common_func_print_just_start(f, name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if self.debug:
print ' printf( "Enter %%s...\\n", "gl%s" );' % (f.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if name not in f.glx_vendorpriv_names:
2005-05-07 18:14:50 +01:00
# XCB specific:
print '#ifdef USE_XCB'
if self.debug:
print ' printf("\\tUsing XCB.\\n");'
print ' xcb_connection_t *c = XGetXCBConnection(dpy);'
2005-05-07 18:14:50 +01:00
print ' (void) __glXFlushRenderBuffer(gc, gc->pc);'
xcb_name = 'xcb_glx%s' % convertStringForXCB(name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
2005-05-07 18:14:50 +01:00
iparams=[]
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
extra_iparams = []
output = None
for p in f.parameterIterator():
if p.is_output:
output = p
if p.is_image():
if p.img_format != "GL_COLOR_INDEX" or p.img_type != "GL_BITMAP":
extra_iparams.append("state->storePack.swapEndian")
else:
extra_iparams.append("0")
# Hardcode this in. lsb_first param (apparently always GL_FALSE)
# also present in GetPolygonStipple, but taken care of above.
if xcb_name == "xcb_glx_read_pixels":
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
extra_iparams.append("0")
else:
2005-05-07 18:14:50 +01:00
iparams.append(p.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
xcb_request = '%s(%s)' % (xcb_name, ", ".join(["c", "gc->currentContextTag"] + iparams + extra_iparams))
2005-05-07 18:14:50 +01:00
if f.needs_reply():
print ' %s_reply_t *reply = %s_reply(c, %s, NULL);' % (xcb_name, xcb_name, xcb_request)
if output:
if output.is_image():
[dim, w, h, d, junk] = output.get_dimensions()
if f.dimensions_in_reply:
w = "reply->width"
h = "reply->height"
d = "reply->depth"
if dim < 2:
h = "1"
else:
print ' if (%s == 0) { %s = 1; }' % (h, h)
if dim < 3:
d = "1"
else:
print ' if (%s == 0) { %s = 1; }' % (d, d)
print ' __glEmptyImage(gc, 3, %s, %s, %s, %s, %s, %s_data(reply), %s);' % (w, h, d, output.img_format, output.img_type, xcb_name, output.name)
else:
if f.reply_always_array:
print ' (void)memcpy(%s, %s_data(reply), %s_data_length(reply) * sizeof(%s));' % (output.name, xcb_name, xcb_name, output.get_base_type_string())
else:
print ' if (%s_data_length(reply) == 0)' % (xcb_name)
print ' (void)memcpy(%s, &reply->datum, sizeof(reply->datum));' % (output.name)
print ' else'
print ' (void)memcpy(%s, %s_data(reply), %s_data_length(reply) * sizeof(%s));' % (output.name, xcb_name, xcb_name, output.get_base_type_string())
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.return_type != 'void':
2005-05-07 18:14:50 +01:00
print ' retval = reply->ret_val;'
print ' free(reply);'
else:
print ' ' + xcb_request + ';'
print '#else'
# End of XCB specific.
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.parameters != []:
pc_decl = "GLubyte const * pc ="
else:
pc_decl = "(void)"
if name in f.glx_vendorpriv_names:
print ' %s __glXSetupVendorRequest(gc, %s, %s, cmdlen);' % (pc_decl, f.opcode_real_name(), f.opcode_vendor_name(name))
else:
print ' %s __glXSetupSingleRequest(gc, %s, cmdlen);' % (pc_decl, f.opcode_name())
self.common_emit_args(f, "pc", 0, 0)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
images = f.get_images()
for img in images:
if img.is_output:
o = f.command_fixed_length() - 4
print ' *(int32_t *)(pc + %u) = 0;' % (o)
if img.img_format != "GL_COLOR_INDEX" or img.img_type != "GL_BITMAP":
print ' * (int8_t *)(pc + %u) = state->storePack.swapEndian;' % (o)
if f.img_reset:
print ' * (int8_t *)(pc + %u) = %s;' % (o + 1, f.img_reset)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return_name = ''
if f.needs_reply():
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.return_type != 'void':
return_name = " retval"
return_str = " retval = (%s)" % (f.return_type)
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return_str = " (void)"
got_reply = 0
for p in f.parameterIterateOutputs():
if p.is_image():
[dim, w, h, d, junk] = p.get_dimensions()
if f.dimensions_in_reply:
print " __glXReadPixelReply(dpy, gc, %u, 0, 0, 0, %s, %s, %s, GL_TRUE);" % (dim, p.img_format, p.img_type, p.name)
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print " __glXReadPixelReply(dpy, gc, %u, %s, %s, %s, %s, %s, %s, GL_FALSE);" % (dim, w, h, d, p.img_format, p.img_type, p.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
got_reply = 1
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.reply_always_array:
aa = "GL_TRUE"
else:
aa = "GL_FALSE"
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
# gl_parameter.size() returns the size
# of the entire data item. If the
# item is a fixed-size array, this is
# the size of the whole array. This
# is not what __glXReadReply wants. It
# wants the size of a single data
# element in the reply packet.
# Dividing by the array size (1 for
# non-arrays) gives us this.
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
s = p.size() / p.get_element_count()
print " %s __glXReadReply(dpy, %s, %s, %s);" % (return_str, s, p.name, aa)
got_reply = 1
# If a reply wasn't read to fill an output parameter,
# read a NULL reply to get the return value.
if not got_reply:
print " %s __glXReadReply(dpy, 0, NULL, GL_FALSE);" % (return_str)
elif self.debug:
# Only emit the extra glFinish call for functions
# that don't already require a reply from the server.
print ' __indirect_glFinish();'
if self.debug:
print ' printf( "Exit %%s.\\n", "gl%s" );' % (name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print ' UnlockDisplay(dpy); SyncHandle();'
if name not in f.glx_vendorpriv_names:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print '#endif /* USE_XCB */'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print ' }'
print ' return%s;' % (return_name)
return
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
def printPixelFunction(self, f):
if self.pixel_stubs.has_key( f.name ):
# Normally gl_function::get_parameter_string could be
# used. However, this call needs to have the missing
# dimensions (e.g., a fake height value for
# glTexImage1D) added in.
p_string = ""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
for param in f.parameterIterateGlxSend():
if param.is_padding:
continue
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
p_string += ", " + param.name
if param.is_image():
[dim, junk, junk, junk, junk] = param.get_dimensions()
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.pad_after(param):
p_string += ", 1"
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
print ' %s(%s, %u%s );' % (self.pixel_stubs[f.name] , f.opcode_name(), dim, p_string)
return
if self.common_func_print_just_start(f, None):
trailer = " }"
else:
trailer = None
if f.can_be_large:
print 'if (cmdlen <= gc->maxSmallRenderCommandSize) {'
print ' if ( (gc->pc + cmdlen) > gc->bufEnd ) {'
print ' (void) __glXFlushRenderBuffer(gc, gc->pc);'
print ' }'
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.glx_rop == ~0:
opcode = "opcode"
else:
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
opcode = f.opcode_real_name()
print 'emit_header(gc->pc, %s, cmdlen);' % (opcode)
self.pixel_emit_args( f, "gc->pc", 0 )
print 'gc->pc += cmdlen;'
print 'if (gc->pc > gc->limit) { (void) __glXFlushRenderBuffer(gc, gc->pc); }'
if f.can_be_large:
print '}'
print 'else {'
self.large_emit_begin(f, opcode)
self.pixel_emit_args(f, "pc", 1)
print '}'
if trailer: print trailer
return
def printRenderFunction(self, f):
# There is a class of GL functions that take a single pointer
# as a parameter. This pointer points to a fixed-size chunk
# of data, and the protocol for this functions is very
# regular. Since they are so regular and there are so many
# of them, special case them with generic functions. On
# x86, this saves about 26KB in the libGL.so binary.
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
if f.variable_length_parameter() == None and len(f.parameters) == 1:
p = f.parameters[0]
if p.is_pointer():
cmdlen = f.command_fixed_length()
if cmdlen in self.generic_sizes:
print ' generic_%u_byte( %s, %s );' % (cmdlen, f.opcode_real_name(), p.name)
return
if self.common_func_print_just_start(f, None):
trailer = " }"
else:
trailer = None
if self.debug:
print 'printf( "Enter %%s...\\n", "gl%s" );' % (f.name)
if f.can_be_large:
print 'if (cmdlen <= gc->maxSmallRenderCommandSize) {'
print ' if ( (gc->pc + cmdlen) > gc->bufEnd ) {'
print ' (void) __glXFlushRenderBuffer(gc, gc->pc);'
print ' }'
print 'emit_header(gc->pc, %s, cmdlen);' % (f.opcode_real_name())
self.common_emit_args(f, "gc->pc", 4, 0)
print 'gc->pc += cmdlen;'
print 'if (__builtin_expect(gc->pc > gc->limit, 0)) { (void) __glXFlushRenderBuffer(gc, gc->pc); }'
if f.can_be_large:
print '}'
print 'else {'
self.large_emit_begin(f)
self.common_emit_args(f, "pc", 8, 1)
p = f.variable_length_parameter()
print ' __glXSendLargeCommand(gc, pc, %u, %s, %s);' % (p.offset + 8, p.name, p.size_string())
print '}'
if self.debug:
print '__indirect_glFinish();'
print 'printf( "Exit %%s.\\n", "gl%s" );' % (f.name)
if trailer: print trailer
return
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
class PrintGlxProtoInit_c(gl_XML.gl_print_base):
def __init__(self):
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
gl_XML.gl_print_base.__init__(self)
self.name = "glX_proto_send.py (from Mesa)"
self.license = license.bsd_license_template % ( \
"""Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
(C) Copyright IBM Corporation 2004""", "PRECISION INSIGHT, IBM")
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
def printRealHeader(self):
print """/**
* \\file indirect_init.c
* Initialize indirect rendering dispatch table.
*
* \\author Kevin E. Martin <kevin@precisioninsight.com>
* \\author Brian Paul <brian@precisioninsight.com>
* \\author Ian Romanick <idr@us.ibm.com>
*/
#include "indirect_init.h"
#include "indirect.h"
#include "glapi.h"
#include <assert.h>
/**
* No-op function used to initialize functions that have no GLX protocol
* support.
*/
static int NoOp(void)
{
return 0;
}
/**
* Create and initialize a new GL dispatch table. The table is initialized
* with GLX indirect rendering protocol functions.
*/
struct _glapi_table * __glXNewIndirectAPI( void )
{
_glapi_proc *table;
unsigned entries;
unsigned i;
int o;
entries = _glapi_get_dispatch_table_size();
table = (_glapi_proc *) malloc(entries * sizeof(_glapi_proc));
/* first, set all entries to point to no-op functions */
for (i = 0; i < entries; i++) {
table[i] = (_glapi_proc) NoOp;
}
/* now, initialize the entries we understand */"""
def printRealFooter(self):
print """
return (struct _glapi_table *) table;
}
"""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
def printBody(self, api):
for [name, number] in api.categoryIterate():
if number != None:
preamble = '\n /* %3u. %s */\n' % (int(number), name)
else:
preamble = '\n /* %s */\n' % (name)
for func in api.functionIterateByCategory(name):
if func.client_supported_for_indirect():
if preamble:
print preamble
preamble = None
if func.is_abi():
print ' table[{offset}] = (_glapi_proc) __indirect_gl{name};'.format(name = func.name, offset = func.offset)
else:
print ' o = _glapi_get_proc_offset("gl{0}");'.format(func.name)
print ' assert(o > 0);'
print ' table[o] = (_glapi_proc) __indirect_gl{0};'.format(func.name)
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
return
class PrintGlxProtoInit_h(gl_XML.gl_print_base):
def __init__(self):
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
gl_XML.gl_print_base.__init__(self)
self.name = "glX_proto_send.py (from Mesa)"
self.license = license.bsd_license_template % ( \
"""Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
(C) Copyright IBM Corporation 2004""", "PRECISION INSIGHT, IBM")
self.header_tag = "_INDIRECT_H_"
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
self.last_category = ""
return
def printRealHeader(self):
print """/**
* \\file
* Prototypes for indirect rendering functions.
*
* \\author Kevin E. Martin <kevin@precisioninsight.com>
* \\author Ian Romanick <idr@us.ibm.com>
*/
"""
self.printFastcall()
self.printNoinline()
print """
#include <X11/Xfuncproto.h>
#include "glxclient.h"
extern _X_HIDDEN NOINLINE CARD32 __glXReadReply( Display *dpy, size_t size,
void * dest, GLboolean reply_is_always_array );
extern _X_HIDDEN NOINLINE void __glXReadPixelReply( Display *dpy,
struct glx_context * gc, unsigned max_dim, GLint width, GLint height,
GLint depth, GLenum format, GLenum type, void * dest,
GLboolean dimensions_in_reply );
extern _X_HIDDEN NOINLINE FASTCALL GLubyte * __glXSetupSingleRequest(
struct glx_context * gc, GLint sop, GLint cmdlen );
extern _X_HIDDEN NOINLINE FASTCALL GLubyte * __glXSetupVendorRequest(
struct glx_context * gc, GLint code, GLint vop, GLint cmdlen );
"""
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
def printBody(self, api):
for func in api.functionIterateGlx():
params = func.get_parameter_string()
print 'extern _X_HIDDEN %s __indirect_gl%s(%s);' % (func.return_type, func.name, params)
for n in func.entry_points:
if func.has_different_protocol(n):
asdf = func.static_glx_name(n)
if asdf not in func.static_entry_points:
print 'extern _X_HIDDEN %s gl%s(%s);' % (func.return_type, asdf, params)
# give it a easy-to-remember name
if func.client_handcode:
print '#define gl_dispatch_stub_%s gl%s' % (n, asdf)
else:
print 'GLAPI %s GLAPIENTRY gl%s(%s);' % (func.return_type, asdf, params)
break
print ''
print '#ifdef GLX_SHARED_GLAPI'
print 'extern _X_HIDDEN void (*__indirect_get_proc_address(const char *name))(void);'
print '#endif'
def show_usage():
print "Usage: %s [-f input_file_name] [-m output_mode] [-d]" % sys.argv[0]
print " -m output_mode Output mode can be one of 'proto', 'init_c' or 'init_h'."
print " -d Enable extra debug information in the generated code."
sys.exit(1)
if __name__ == '__main__':
file_name = "gl_API.xml"
try:
(args, trail) = getopt.getopt(sys.argv[1:], "f:m:d")
except Exception,e:
show_usage()
debug = 0
mode = "proto"
for (arg,val) in args:
if arg == "-f":
file_name = val
elif arg == "-m":
mode = val
elif arg == "-d":
debug = 1
if mode == "proto":
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
printer = PrintGlxProtoStubs()
elif mode == "init_c":
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
printer = PrintGlxProtoInit_c()
elif mode == "init_h":
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
printer = PrintGlxProtoInit_h()
else:
show_usage()
Mammoth update to the Python code generator scripts that live in src/mesa/glapi. Basically, the scripts that did simple things (like gl_offsets.py) were simple, and the scripts that did more complicated things (like glX_proto_send.py) were getting progressively more and more out of control. So, I re-write the foundation classes on which everything is based. One problem with the existing code is that the division between the GL API database representation and the way the output code is generated was either blury or nonexistant. The new code somewhat follows the Model-View-Controller pattern, minus the Controller. There is a distinct set of classes that model the API data, and there is a distinct set of classes that generate code from that data. One big change is in the class that represents GL functions (was glFunction, is now gl_function). There used to be an instance of this calls for each function and for each alias to that function. For example, there was an instance for PointParameterivSGIS, PointParameterivEXT, PointParameterivARB, and PointParameteriv. In the new code, there is one instance. Each instance has a list of entrypoint names for the function. In the next revision, this will allow a couple useful things. The script will be able to verify that the parameters, return type, and GLX protocol for a function and all it's aliases match. It will also allow aliases to be represented in the XML more compactly. Instead of repeating all the information, an alias can be listed as: <function name="PointParameterivARB" alias="PointParameterivEXT"/> Because the data representation was changed, the order that the alias functions are processed by the scripts also changed. This accounts for at least 2,700 of the ~3,600 lines of diffs in the generated code. Most of the remaining ~900 lines of diffs are the result of bugs *fixed* by the new scripts. The old scripts also generated code with some bugs in it. These bugs were discovered while the new code was being written. These changes were discussed on the mesa3d-dev mailing list back at the end of May: http://marc.theaimsgroup.com/?t=111714569000004&r=1&w=2 Xorg bug: 3197, 3208
2005-06-22 00:42:43 +01:00
printer.debug = debug
api = gl_XML.parse_GL_API( file_name, glX_XML.glx_item_factory() )
printer.Print( api )