Vulkan-Docs/src/spec/cgenerator.py

346 lines
16 KiB
Python
Raw Normal View History

#!/usr/bin/python3 -i
#
Change log for January 23, 2017 Vulkan 1.0.39 spec update: * Bump API patch number and header version number to 39 for this update. Github Issues: * Clarified that only accesses via the specified buffer/image subresource ranges are included in the access scopes (public issue 306). * Add missing valid usage statements for flink:vkCreateComputePipelines and flink:vkCreateGraphicsPipelines (public issue 427). Internal Issues: * Add a Note to the <<invariance,Invariance>> appendix about a difference between OpenGL and Vulkan with regards to how primitives derived from offsets are handled (internal issue 355). * Add the +<<VK_KHR_get_physical_device_properties2>>+, +<<VK_KHR_maintenance1>>+, and +<<VK_KHR_shader_draw_parameters>>+ extensions (internal issue 448). * Add the +<<VK_EXT_shader_subgroup_vote>>+ and +<<VK_EXT_shader_subgroup_ballot>>+ extensions (internal issue 449). * Update the texture level-of-detail equation in the <<textures-scale-factor,Scale Factor Operation>> section to better approximate the ellipse major and minor axes (internal issue 547). * Forbid non-explicitly allowed uses of interface decorations in the introduction to the <<interfaces,Shader Interfaces>> chapter (internal issue 607). * Replace use of MathJax with KaTeX, for improved load-time performance as well as avoiding the scrolling-and-scrolling behavior due to MathJax asynchronous rendering when loading at an anchor inside the spec. This change also requires moving to HTML5 output for the spec instead of XHTML, and there is a visible difference in that the chapter navigation index is now in a scrollable sidebar instead of at the top of the document. We may or may not retain the nav sidebar based on feedback (internal issue 613). * Improve consistency of markup and formatting in extension appendices (internal issue 631). Other Issues: * Add explicit valid usage statements to slink:VkImageCopy requiring that the source and destination layer ranges be contained in their respective source and destination images. * Add valid usage language for swapchain of flink:vkAcquireNextImage. If the swapchain has been replaced, then it should not be passed to flink:vkAcquireNextImage. * Add a valid usage statement to flink:vkCreateImageView, that the image must have been created with an appropriate usage bit set. * Noted that slink:VkDisplayPresentInfoKHR is a valid extension of slink:VkPresentInfoKHR in the <<wsi_swapchain,WSI Swapchain>> section. * Update valid usage for flink:vkCmdSetViewport and flink:vkCmdSetScissor to account for the multiple viewport feature. If the feature is not enabled, the parameters for these functions have required values that are defined in the <<features-features-multiViewport,multiple viewports>> section of the spec but were not reflected in the valid usage text for these functions. * Add the +<<VK_EXT_swapchain_colorspace>>+ extension defining common color spaces.
2017-01-18 04:11:25 +00:00
# Copyright (c) 2013-2017 The Khronos Group Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os,re,sys
from generator import *
# CGeneratorOptions - subclass of GeneratorOptions.
#
# Adds options used by COutputGenerator objects during C language header
# generation.
#
# Additional members
# prefixText - list of strings to prefix generated header with
# (usually a copyright statement + calling convention macros).
# protectFile - True if multiple inclusion protection should be
# generated (based on the filename) around the entire header.
# protectFeature - True if #ifndef..#endif protection should be
# generated around a feature interface in the header file.
# genFuncPointers - True if function pointer typedefs should be
# generated
# protectProto - If conditional protection should be generated
# around prototype declarations, set to either '#ifdef'
# to require opt-in (#ifdef protectProtoStr) or '#ifndef'
# to require opt-out (#ifndef protectProtoStr). Otherwise
# set to None.
# protectProtoStr - #ifdef/#ifndef symbol to use around prototype
# declarations, if protectProto is set
# apicall - string to use for the function declaration prefix,
# such as APICALL on Windows.
# apientry - string to use for the calling convention macro,
# in typedefs, such as APIENTRY.
# apientryp - string to use for the calling convention macro
# in function pointer typedefs, such as APIENTRYP.
# indentFuncProto - True if prototype declarations should put each
# parameter on a separate line
# indentFuncPointer - True if typedefed function pointers should put each
# parameter on a separate line
# alignFuncParam - if nonzero and parameters are being put on a
# separate line, align parameter names at the specified column
class CGeneratorOptions(GeneratorOptions):
"""Represents options during C interface generation for headers"""
def __init__(self,
filename = None,
Change log for July 22, 2016 Vulkan 1.0.22 spec update: * Bump API patch number and header version number to 22 for this update. Github Issues: * Translate the subpass self-dependency language into concrete validity statements, and added a validity statement about the restrictions on layout parameters (public issue 267). * Add validity requirement that slink:VkAttachmentDescription::pname:finalLayout and slink:VkAttachmentReference::pname:layout must not be ename:VK_IMAGE_LAYOUT_UNDEFINED or ename:VK_IMAGE_LAYOUT_PREINITIALIZED (public issue 268). * Clarify that slink:VkSubpassDescription::pname:pResolveAttachments layouts are used. Make language consistent with other attachment arrays (public issue 270). * Changed 64-bit definition for dname:VK_DEFINE_NON_DISPATCHABLE_HANDLE to work for x32 platform in +vk.xml+ and the resulting +vulkan.h+ (public issue 282). * Add missing error return code for flink:vkEnumerateInstanceExtensionProperties and flink:vkEnumerateDeviceExtensionProperties (public issue 285) * Fix several cases of stext::VkStructName.memberName markup to stext::VkStructName::pname:memberName, to match other usage in the spec, and describe this markup in the style guide (public issue 286). * Modified validity language generation script to avoid redundant common ancestor language if covered by generic parent language, and used `Both' instead of `Each' when appropriate (public issue 288). Internal Issues: * Add language about behavior of flink:vkAllocateDescriptorSets when allocation fails due to fragmentation, a new error ename:VK_ERROR_FRAGMENTED_POOL, and a Note explaining the situation (internal issue 309). * For the features of code:PointSize, code:ClipDistance, and code:CullDistance, the SPIR-V capability is required to be declared on use (read or write) rather than on decoration (internal issue 359). * Have desktop versions of GLSL respect precision qualification (code:mediump and code:lowp) when compiling for Vulkan. These will get translated to SPIR-V's code:RelaxedPrecision decoration as they do with OpenGL ES versions of GLSL (ESSL). The default precision of all types is code:highp when using a desktop version (internal issue 360). * Add validity statement for slink:VkImageCreateInfo specifying that multisampled images must be two-dimensional, optimally tiled, and with a single mipmap level (internal issue 369). * Add validity statements to slink:VkImageViewCreateInfo disallowing creation of images or image views with no supported features. Made some slink:VkImageViewCreateInfo validity statements more precise and consistent. Added a Note to the <<features,features>> chapter about formats with no features (internal issue 371). * Remove +manpages+ from default build targets. Nroff outputs containing imbedded latexmath will not render properly. Fixing this is a lot of work for limited use cases (internal issue 401). Other Commits: * Fix flink:vkRenderPassBeginInfo::pname:clearValueCount validity statement to be based on attachment indices rather than the number of cleared attachments (Vulkan-LoaderAndValidationLayers/issues/601). * Convert registry documentation from LaTeX to asciidoc source and rename from +src/spec/readme.tex+ to +src/spec/registry.txt+. * Fix lack of Oxford commas in validity language. * Lots of cleanup of generator scripts and Makefiles to move extension list for generator into the script arguments instead of the body of genvk.py, and express better dependencies between XML, scripts, and generated files.
2016-07-23 10:15:48 +00:00
directory = '.',
apiname = None,
profile = None,
versions = '.*',
emitversions = '.*',
defaultExtensions = None,
addExtensions = None,
removeExtensions = None,
sortProcedure = regSortFeatures,
prefixText = "",
genFuncPointers = True,
protectFile = True,
protectFeature = True,
protectProto = None,
protectProtoStr = None,
apicall = '',
apientry = '',
apientryp = '',
indentFuncProto = True,
indentFuncPointer = False,
alignFuncParam = 0):
Change log for July 22, 2016 Vulkan 1.0.22 spec update: * Bump API patch number and header version number to 22 for this update. Github Issues: * Translate the subpass self-dependency language into concrete validity statements, and added a validity statement about the restrictions on layout parameters (public issue 267). * Add validity requirement that slink:VkAttachmentDescription::pname:finalLayout and slink:VkAttachmentReference::pname:layout must not be ename:VK_IMAGE_LAYOUT_UNDEFINED or ename:VK_IMAGE_LAYOUT_PREINITIALIZED (public issue 268). * Clarify that slink:VkSubpassDescription::pname:pResolveAttachments layouts are used. Make language consistent with other attachment arrays (public issue 270). * Changed 64-bit definition for dname:VK_DEFINE_NON_DISPATCHABLE_HANDLE to work for x32 platform in +vk.xml+ and the resulting +vulkan.h+ (public issue 282). * Add missing error return code for flink:vkEnumerateInstanceExtensionProperties and flink:vkEnumerateDeviceExtensionProperties (public issue 285) * Fix several cases of stext::VkStructName.memberName markup to stext::VkStructName::pname:memberName, to match other usage in the spec, and describe this markup in the style guide (public issue 286). * Modified validity language generation script to avoid redundant common ancestor language if covered by generic parent language, and used `Both' instead of `Each' when appropriate (public issue 288). Internal Issues: * Add language about behavior of flink:vkAllocateDescriptorSets when allocation fails due to fragmentation, a new error ename:VK_ERROR_FRAGMENTED_POOL, and a Note explaining the situation (internal issue 309). * For the features of code:PointSize, code:ClipDistance, and code:CullDistance, the SPIR-V capability is required to be declared on use (read or write) rather than on decoration (internal issue 359). * Have desktop versions of GLSL respect precision qualification (code:mediump and code:lowp) when compiling for Vulkan. These will get translated to SPIR-V's code:RelaxedPrecision decoration as they do with OpenGL ES versions of GLSL (ESSL). The default precision of all types is code:highp when using a desktop version (internal issue 360). * Add validity statement for slink:VkImageCreateInfo specifying that multisampled images must be two-dimensional, optimally tiled, and with a single mipmap level (internal issue 369). * Add validity statements to slink:VkImageViewCreateInfo disallowing creation of images or image views with no supported features. Made some slink:VkImageViewCreateInfo validity statements more precise and consistent. Added a Note to the <<features,features>> chapter about formats with no features (internal issue 371). * Remove +manpages+ from default build targets. Nroff outputs containing imbedded latexmath will not render properly. Fixing this is a lot of work for limited use cases (internal issue 401). Other Commits: * Fix flink:vkRenderPassBeginInfo::pname:clearValueCount validity statement to be based on attachment indices rather than the number of cleared attachments (Vulkan-LoaderAndValidationLayers/issues/601). * Convert registry documentation from LaTeX to asciidoc source and rename from +src/spec/readme.tex+ to +src/spec/registry.txt+. * Fix lack of Oxford commas in validity language. * Lots of cleanup of generator scripts and Makefiles to move extension list for generator into the script arguments instead of the body of genvk.py, and express better dependencies between XML, scripts, and generated files.
2016-07-23 10:15:48 +00:00
GeneratorOptions.__init__(self, filename, directory, apiname, profile,
versions, emitversions, defaultExtensions,
addExtensions, removeExtensions, sortProcedure)
self.prefixText = prefixText
self.genFuncPointers = genFuncPointers
self.protectFile = protectFile
self.protectFeature = protectFeature
self.protectProto = protectProto
self.protectProtoStr = protectProtoStr
self.apicall = apicall
self.apientry = apientry
self.apientryp = apientryp
self.indentFuncProto = indentFuncProto
self.indentFuncPointer = indentFuncPointer
self.alignFuncParam = alignFuncParam
# COutputGenerator - subclass of OutputGenerator.
# Generates C-language API interfaces.
#
# ---- methods ----
# COutputGenerator(errFile, warnFile, diagFile) - args as for
# OutputGenerator. Defines additional internal state.
# ---- methods overriding base class ----
# beginFile(genOpts)
# endFile()
# beginFeature(interface, emit)
# endFeature()
# genType(typeinfo,name)
# genStruct(typeinfo,name)
# genGroup(groupinfo,name)
# genEnum(enuminfo, name)
# genCmd(cmdinfo)
class COutputGenerator(OutputGenerator):
"""Generate specified API interfaces in a specific style, such as a C header"""
# This is an ordered list of sections in the header file.
TYPE_SECTIONS = ['include', 'define', 'basetype', 'handle', 'enum',
'group', 'bitmask', 'funcpointer', 'struct']
ALL_SECTIONS = TYPE_SECTIONS + ['commandPointer', 'command']
def __init__(self,
errFile = sys.stderr,
warnFile = sys.stderr,
diagFile = sys.stdout):
OutputGenerator.__init__(self, errFile, warnFile, diagFile)
# Internal state - accumulators for different inner block text
self.sections = dict([(section, []) for section in self.ALL_SECTIONS])
#
def beginFile(self, genOpts):
OutputGenerator.beginFile(self, genOpts)
# C-specific
#
# Multiple inclusion protection & C++ wrappers.
if (genOpts.protectFile and self.genOpts.filename):
headerSym = re.sub('\.h', '_h_',
os.path.basename(self.genOpts.filename)).upper()
write('#ifndef', headerSym, file=self.outFile)
write('#define', headerSym, '1', file=self.outFile)
self.newline()
write('#ifdef __cplusplus', file=self.outFile)
write('extern "C" {', file=self.outFile)
write('#endif', file=self.outFile)
self.newline()
#
# User-supplied prefix text, if any (list of strings)
if (genOpts.prefixText):
for s in genOpts.prefixText:
write(s, file=self.outFile)
#
# Some boilerplate describing what was generated - this
# will probably be removed later since the extensions
# pattern may be very long.
# write('/* Generated C header for:', file=self.outFile)
# write(' * API:', genOpts.apiname, file=self.outFile)
# if (genOpts.profile):
# write(' * Profile:', genOpts.profile, file=self.outFile)
# write(' * Versions considered:', genOpts.versions, file=self.outFile)
# write(' * Versions emitted:', genOpts.emitversions, file=self.outFile)
# write(' * Default extensions included:', genOpts.defaultExtensions, file=self.outFile)
# write(' * Additional extensions included:', genOpts.addExtensions, file=self.outFile)
# write(' * Extensions removed:', genOpts.removeExtensions, file=self.outFile)
# write(' */', file=self.outFile)
def endFile(self):
# C-specific
# Finish C++ wrapper and multiple inclusion protection
self.newline()
write('#ifdef __cplusplus', file=self.outFile)
write('}', file=self.outFile)
write('#endif', file=self.outFile)
if (self.genOpts.protectFile and self.genOpts.filename):
self.newline()
write('#endif', file=self.outFile)
# Finish processing in superclass
OutputGenerator.endFile(self)
def beginFeature(self, interface, emit):
# Start processing in superclass
OutputGenerator.beginFeature(self, interface, emit)
# C-specific
# Accumulate includes, defines, types, enums, function pointer typedefs,
# end function prototypes separately for this feature. They're only
# printed in endFeature().
self.sections = dict([(section, []) for section in self.ALL_SECTIONS])
def endFeature(self):
# C-specific
# Actually write the interface to the output file.
if (self.emit):
self.newline()
if (self.genOpts.protectFeature):
write('#ifndef', self.featureName, file=self.outFile)
# If type declarations are needed by other features based on
# this one, it may be necessary to suppress the ExtraProtect,
# or move it below the 'for section...' loop.
if (self.featureExtraProtect != None):
write('#ifdef', self.featureExtraProtect, file=self.outFile)
write('#define', self.featureName, '1', file=self.outFile)
for section in self.TYPE_SECTIONS:
contents = self.sections[section]
if contents:
write('\n'.join(contents), file=self.outFile)
self.newline()
if (self.genOpts.genFuncPointers and self.sections['commandPointer']):
write('\n'.join(self.sections['commandPointer']), file=self.outFile)
self.newline()
if (self.sections['command']):
if (self.genOpts.protectProto):
write(self.genOpts.protectProto,
self.genOpts.protectProtoStr, file=self.outFile)
write('\n'.join(self.sections['command']), end='', file=self.outFile)
if (self.genOpts.protectProto):
write('#endif', file=self.outFile)
else:
self.newline()
if (self.featureExtraProtect != None):
write('#endif /*', self.featureExtraProtect, '*/', file=self.outFile)
if (self.genOpts.protectFeature):
write('#endif /*', self.featureName, '*/', file=self.outFile)
# Finish processing in superclass
OutputGenerator.endFeature(self)
#
# Append a definition to the specified section
def appendSection(self, section, text):
# self.sections[section].append('SECTION: ' + section + '\n')
self.sections[section].append(text)
#
# Type generation
def genType(self, typeinfo, name):
OutputGenerator.genType(self, typeinfo, name)
typeElem = typeinfo.elem
# If the type is a struct type, traverse the imbedded <member> tags
# generating a structure. Otherwise, emit the tag text.
category = typeElem.get('category')
if (category == 'struct' or category == 'union'):
self.genStruct(typeinfo, name)
else:
# Replace <apientry /> tags with an APIENTRY-style string
# (from self.genOpts). Copy other text through unchanged.
# If the resulting text is an empty string, don't emit it.
s = noneStr(typeElem.text)
for elem in typeElem:
if (elem.tag == 'apientry'):
s += self.genOpts.apientry + noneStr(elem.tail)
else:
s += noneStr(elem.text) + noneStr(elem.tail)
if s:
# Add extra newline after multi-line entries.
if '\n' in s:
s += '\n'
self.appendSection(category, s)
#
# Struct (e.g. C "struct" type) generation.
# This is a special case of the <type> tag where the contents are
# interpreted as a set of <member> tags instead of freeform C
# C type declarations. The <member> tags are just like <param>
# tags - they are a declaration of a struct or union member.
# Only simple member declarations are supported (no nested
# structs etc.)
def genStruct(self, typeinfo, typeName):
OutputGenerator.genStruct(self, typeinfo, typeName)
body = 'typedef ' + typeinfo.elem.get('category') + ' ' + typeName + ' {\n'
# paramdecl = self.makeCParamDecl(typeinfo.elem, self.genOpts.alignFuncParam)
targetLen = 0;
for member in typeinfo.elem.findall('.//member'):
targetLen = max(targetLen, self.getCParamTypeLength(member))
for member in typeinfo.elem.findall('.//member'):
body += self.makeCParamDecl(member, targetLen + 4)
body += ';\n'
body += '} ' + typeName + ';\n'
self.appendSection('struct', body)
#
# Group (e.g. C "enum" type) generation.
# These are concatenated together with other types.
def genGroup(self, groupinfo, groupName):
OutputGenerator.genGroup(self, groupinfo, groupName)
groupElem = groupinfo.elem
expandName = re.sub(r'([0-9a-z_])([A-Z0-9][^A-Z0-9]?)',r'\1_\2',groupName).upper()
expandPrefix = expandName
expandSuffix = ''
expandSuffixMatch = re.search(r'[A-Z][A-Z]+$',groupName)
if expandSuffixMatch:
expandSuffix = '_' + expandSuffixMatch.group()
# Strip off the suffix from the prefix
expandPrefix = expandName.rsplit(expandSuffix, 1)[0]
# Prefix
body = "\ntypedef enum " + groupName + " {\n"
Change log for August 26, 2016 Vulkan 1.0.25 spec update: * Bump API patch number and header version number to 25 for this update. * Structurally change the specification so that multiple extensions are included in the +1.0+ git branch, and specifications will include or not include those extensions at build time based on options passed to the Makefile. See +doc/specs/vulkan/README.html+ and the ``Layers and Extensions'' section of the ``Vulkan Documentation and Extensions'' document for more information on this change. * Register and publish new extensions in the single-branch form: ** +VK_NV_external_memory_capabilities+ ** +VK_NV_external_memory+ ** +VK_NV_external_memory_win32+ ** +VK_NV_win32_keyed_mutex+ Github Issues: * Clarify description of GetInstanceProcAddr and GetDeviceProcAddr (public issue 212). * Add SPIR-V <<textures-operation-validation, instruction validation>> for single-sampled images (public issue 316). * Fix spelling of ``tesselation'' in a few places and note it as an exception to the American spelling rules convention (public issue 327). * Fix Makefile to create output directory for ``styleguide'' target (public issue 329). * Fix numerous minor issues with incorrectly tags on enumerant names and table titles (public issue 330). * Generate specversion.txt date in UTC time and RFC 2822 format (public issue 335). * Convert link to the SPIR-V Specification for flink:VkShaderModuleCreateInfo into an internal link to the normative reference (public issue 336). * Add ename:VK_ERROR_OUT_OF_MEMORY error code to flink:vkCreateDebugReportCallbackEXT (public issue 337). Internal Issues: * Update style guide regarding use of code:NULL and dname:VK_NULL_HANDLE (internal issue 393). * Change the definition of latexmath:[$q$] in the <<textures-image-level-selection,texture image level selection>> section to be the index of the maximum defined level for the view, not the number of levels in the view (internal issue 406). * Allow developers to override dname:VK_DEFINE_NON_DISPATCHABLE_HANDLE with their own binary-compatible definition (internal issue 439). * Fix +vk_platform.h+ conditional logic causing compile failure with some Android compilers (internal issue 441). * Implement the single-branch model as described above (internal issue 461). Other Issues:
2016-08-28 10:47:19 +00:00
# @@ Should use the type="bitmask" attribute instead
isEnum = ('FLAG_BITS' not in expandPrefix)
# Loop over the nested 'enum' tags. Keep track of the minimum and
# maximum numeric values, if they can be determined; but only for
# core API enumerants, not extension enumerants. This is inferred
# by looking for 'extends' attributes.
minName = None
for elem in groupElem.findall('enum'):
# Convert the value to an integer and use that to track min/max.
# Values of form -(number) are accepted but nothing more complex.
# Should catch exceptions here for more complex constructs. Not yet.
(numVal,strVal) = self.enumToValue(elem, True)
name = elem.get('name')
Change log for August 26, 2016 Vulkan 1.0.25 spec update: * Bump API patch number and header version number to 25 for this update. * Structurally change the specification so that multiple extensions are included in the +1.0+ git branch, and specifications will include or not include those extensions at build time based on options passed to the Makefile. See +doc/specs/vulkan/README.html+ and the ``Layers and Extensions'' section of the ``Vulkan Documentation and Extensions'' document for more information on this change. * Register and publish new extensions in the single-branch form: ** +VK_NV_external_memory_capabilities+ ** +VK_NV_external_memory+ ** +VK_NV_external_memory_win32+ ** +VK_NV_win32_keyed_mutex+ Github Issues: * Clarify description of GetInstanceProcAddr and GetDeviceProcAddr (public issue 212). * Add SPIR-V <<textures-operation-validation, instruction validation>> for single-sampled images (public issue 316). * Fix spelling of ``tesselation'' in a few places and note it as an exception to the American spelling rules convention (public issue 327). * Fix Makefile to create output directory for ``styleguide'' target (public issue 329). * Fix numerous minor issues with incorrectly tags on enumerant names and table titles (public issue 330). * Generate specversion.txt date in UTC time and RFC 2822 format (public issue 335). * Convert link to the SPIR-V Specification for flink:VkShaderModuleCreateInfo into an internal link to the normative reference (public issue 336). * Add ename:VK_ERROR_OUT_OF_MEMORY error code to flink:vkCreateDebugReportCallbackEXT (public issue 337). Internal Issues: * Update style guide regarding use of code:NULL and dname:VK_NULL_HANDLE (internal issue 393). * Change the definition of latexmath:[$q$] in the <<textures-image-level-selection,texture image level selection>> section to be the index of the maximum defined level for the view, not the number of levels in the view (internal issue 406). * Allow developers to override dname:VK_DEFINE_NON_DISPATCHABLE_HANDLE with their own binary-compatible definition (internal issue 439). * Fix +vk_platform.h+ conditional logic causing compile failure with some Android compilers (internal issue 441). * Implement the single-branch model as described above (internal issue 461). Other Issues:
2016-08-28 10:47:19 +00:00
# Extension enumerants are only included if they are required
if (self.isEnumRequired(elem)):
body += " " + name + " = " + strVal + ",\n"
Change log for August 26, 2016 Vulkan 1.0.25 spec update: * Bump API patch number and header version number to 25 for this update. * Structurally change the specification so that multiple extensions are included in the +1.0+ git branch, and specifications will include or not include those extensions at build time based on options passed to the Makefile. See +doc/specs/vulkan/README.html+ and the ``Layers and Extensions'' section of the ``Vulkan Documentation and Extensions'' document for more information on this change. * Register and publish new extensions in the single-branch form: ** +VK_NV_external_memory_capabilities+ ** +VK_NV_external_memory+ ** +VK_NV_external_memory_win32+ ** +VK_NV_win32_keyed_mutex+ Github Issues: * Clarify description of GetInstanceProcAddr and GetDeviceProcAddr (public issue 212). * Add SPIR-V <<textures-operation-validation, instruction validation>> for single-sampled images (public issue 316). * Fix spelling of ``tesselation'' in a few places and note it as an exception to the American spelling rules convention (public issue 327). * Fix Makefile to create output directory for ``styleguide'' target (public issue 329). * Fix numerous minor issues with incorrectly tags on enumerant names and table titles (public issue 330). * Generate specversion.txt date in UTC time and RFC 2822 format (public issue 335). * Convert link to the SPIR-V Specification for flink:VkShaderModuleCreateInfo into an internal link to the normative reference (public issue 336). * Add ename:VK_ERROR_OUT_OF_MEMORY error code to flink:vkCreateDebugReportCallbackEXT (public issue 337). Internal Issues: * Update style guide regarding use of code:NULL and dname:VK_NULL_HANDLE (internal issue 393). * Change the definition of latexmath:[$q$] in the <<textures-image-level-selection,texture image level selection>> section to be the index of the maximum defined level for the view, not the number of levels in the view (internal issue 406). * Allow developers to override dname:VK_DEFINE_NON_DISPATCHABLE_HANDLE with their own binary-compatible definition (internal issue 439). * Fix +vk_platform.h+ conditional logic causing compile failure with some Android compilers (internal issue 441). * Implement the single-branch model as described above (internal issue 461). Other Issues:
2016-08-28 10:47:19 +00:00
if (isEnum and elem.get('extends') is None):
if (minName == None):
minName = maxName = name
minValue = maxValue = numVal
elif (numVal < minValue):
minName = name
minValue = numVal
elif (numVal > maxValue):
maxName = name
maxValue = numVal
# Generate min/max value tokens and a range-padding enum. Need some
# additional padding to generate correct names...
if isEnum:
body += " " + expandPrefix + "_BEGIN_RANGE" + expandSuffix + " = " + minName + ",\n"
body += " " + expandPrefix + "_END_RANGE" + expandSuffix + " = " + maxName + ",\n"
body += " " + expandPrefix + "_RANGE_SIZE" + expandSuffix + " = (" + maxName + " - " + minName + " + 1),\n"
body += " " + expandPrefix + "_MAX_ENUM" + expandSuffix + " = 0x7FFFFFFF\n"
# Postfix
body += "} " + groupName + ";"
if groupElem.get('type') == 'bitmask':
section = 'bitmask'
else:
section = 'group'
self.appendSection(section, body)
# Enumerant generation
# <enum> tags may specify their values in several ways, but are usually
# just integers.
def genEnum(self, enuminfo, name):
OutputGenerator.genEnum(self, enuminfo, name)
(numVal,strVal) = self.enumToValue(enuminfo.elem, False)
body = '#define ' + name.ljust(33) + ' ' + strVal
self.appendSection('enum', body)
#
# Command generation
def genCmd(self, cmdinfo, name):
OutputGenerator.genCmd(self, cmdinfo, name)
#
decls = self.makeCDecls(cmdinfo.elem)
self.appendSection('command', decls[0] + '\n')
if (self.genOpts.genFuncPointers):
self.appendSection('commandPointer', decls[1])