Multiple equivalent require blocks could be merged

[Waffl3x]

I am fairly certain the former can be replaced with the latter as the require blocks are equivalent, only differing in their depends attribute. I can open a pull request for this later if this change is desirable.

        <extension name="VK_KHR_push_descriptor" number="81" type="device" author="KHR" depends="VK_KHR_get_physical_device_properties2,VK_VERSION_1_1" contact="Jeff Bolz @jeffbolznv" supported="vulkan" ratified="vulkan">
            <require>
                <enum value="2"                                             name="VK_KHR_PUSH_DESCRIPTOR_SPEC_VERSION"/>
                <enum value="&quot;VK_KHR_push_descriptor&quot;"            name="VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME"/>
                <enum offset="0" extends="VkStructureType"                  name="VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR"/>
                <enum bitpos="0" extends="VkDescriptorSetLayoutCreateFlagBits"   name="VK_DESCRIPTOR_SET_LAYOUT_CREATE_PUSH_DESCRIPTOR_BIT_KHR"  comment="Descriptors are pushed via flink:vkCmdPushDescriptorSetKHR"/>
                <command name="vkCmdPushDescriptorSetKHR"/>
                <type name="VkPhysicalDevicePushDescriptorPropertiesKHR"/>
            </require>
            <require depends="VK_VERSION_1_1">
                <command name="vkCmdPushDescriptorSetWithTemplateKHR"/>
                <enum value="1" extends="VkDescriptorUpdateTemplateType"    name="VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_PUSH_DESCRIPTORS_KHR" comment="Create descriptor update template for pushed descriptor updates"/>
            </require>
            <require depends="VK_KHR_descriptor_update_template">
                <command name="vkCmdPushDescriptorSetWithTemplateKHR"/>
                <enum value="1" extends="VkDescriptorUpdateTemplateType"    name="VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_PUSH_DESCRIPTORS_KHR" comment="Create descriptor update template for pushed descriptor updates"/>
            </require>
        </extension>

        <extension name="VK_KHR_push_descriptor" number="81" type="device" author="KHR" depends="VK_KHR_get_physical_device_properties2,VK_VERSION_1_1" contact="Jeff Bolz @jeffbolznv" supported="vulkan" ratified="vulkan">
            <require>
                <enum value="2"                                             name="VK_KHR_PUSH_DESCRIPTOR_SPEC_VERSION"/>
                <enum value="&quot;VK_KHR_push_descriptor&quot;"            name="VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME"/>
                <enum offset="0" extends="VkStructureType"                  name="VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR"/>
                <enum bitpos="0" extends="VkDescriptorSetLayoutCreateFlagBits"   name="VK_DESCRIPTOR_SET_LAYOUT_CREATE_PUSH_DESCRIPTOR_BIT_KHR"  comment="Descriptors are pushed via flink:vkCmdPushDescriptorSetKHR"/>
                <command name="vkCmdPushDescriptorSetKHR"/>
                <type name="VkPhysicalDevicePushDescriptorPropertiesKHR"/>
            </require>
            <require depends="VK_VERSION_1_1,VK_KHR_descriptor_update_template">
                <command name="vkCmdPushDescriptorSetWithTemplateKHR"/>
                <enum value="1" extends="VkDescriptorUpdateTemplateType"    name="VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_PUSH_DESCRIPTORS_KHR" comment="Create descriptor update template for pushed descriptor updates"/>
            </require>
        </extension>

[Waffl3x]

I found another example, I'll probably have to write a script to locate all cases.

        <extension name="VK_EXT_full_screen_exclusive" number="256" type="device" author="EXT" depends="(VK_KHR_get_physical_device_properties2,VK_VERSION_1_1)+VK_KHR_surface+VK_KHR_get_surface_capabilities2+VK_KHR_swapchain" platform="win32" contact="James Jones @cubanismo" supported="vulkan" ratified="vulkan">
            <require>
                <enum value="4"                                             name="VK_EXT_FULL_SCREEN_EXCLUSIVE_SPEC_VERSION"/>
                <enum value="&quot;VK_EXT_full_screen_exclusive&quot;"      name="VK_EXT_FULL_SCREEN_EXCLUSIVE_EXTENSION_NAME"/>
                <enum offset="0" extends="VkStructureType"                  name="VK_STRUCTURE_TYPE_SURFACE_FULL_SCREEN_EXCLUSIVE_INFO_EXT"/>
                <enum offset="2" extends="VkStructureType"                  name="VK_STRUCTURE_TYPE_SURFACE_CAPABILITIES_FULL_SCREEN_EXCLUSIVE_EXT"/>
                <enum offset="0" extends="VkResult" dir="-"                 name="VK_ERROR_FULL_SCREEN_EXCLUSIVE_MODE_LOST_EXT"/>
                <type name="VkFullScreenExclusiveEXT"/>
                <type name="VkSurfaceFullScreenExclusiveInfoEXT"/>
                <type name="VkSurfaceCapabilitiesFullScreenExclusiveEXT"/>
                <command name="vkGetPhysicalDeviceSurfacePresentModes2EXT"/>
                <command name="vkAcquireFullScreenExclusiveModeEXT"/>
                <command name="vkReleaseFullScreenExclusiveModeEXT"/>
            </require>
            <require depends="VK_KHR_win32_surface">
                <enum offset="1" extends="VkStructureType"                  name="VK_STRUCTURE_TYPE_SURFACE_FULL_SCREEN_EXCLUSIVE_WIN32_INFO_EXT"/>
                <type name="VkSurfaceFullScreenExclusiveWin32InfoEXT"/>
            </require>
            <require depends="VK_KHR_device_group">
                <command name="vkGetDeviceGroupSurfacePresentModes2EXT"/>
            </require>
            <require depends="VK_VERSION_1_1">
                <command name="vkGetDeviceGroupSurfacePresentModes2EXT"/>
            </require>
        </extension>

[oddhack]

Agreed this is harmless in terms of the generated artifacts from the spec repo. We'll discuss as to whether there might be any downstream consequences.

Sometimes we refactor interactions like this from one extension to another, which is probably how this happened.

[Waffl3x]

Sometimes we refactor interactions like this from one extension to another, which is probably how this happened.

The theory I heard was it happened during the transition from requires and requiresCore to depends. Sounds very plausible to me as this is how it would have been structured for those attributes.

[oddhack]

We discussed and are OK with this - please go ahead and propose a PR to that effect. If you happen to write a detection tool we could run in CI, that seems useful as well (preferably as a Python script).

[oddhack]

Closed with #2406 merged and published in the latest spec update.