| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
Compile include/ headers on their own In https://review.skia.org/1024156, we removed IWYU logic, but this also removed the ability for us to test compiling headers on their own (which other build systems do). This caused a missing #include to land in Skia but break upstream things [1]. By adding in the way to generate .cpp files for headers, this will allow us to catch that type of error. Unlike before, where there was a hand-curated list of headers, I did my best to include all public headers (and then those from select/core) which would have caught the previous error. If we like this, I can add other subdirectories to src to the list. I also did my best to avoid duplication of header files, so they are listed in one place and not two (e.g. 2 different BUILD.bazel rules) so it's hopefully easier to maintain. Change-Id: Ib5589c6ac06709d5a472bc7c524cc7a3b2a506b0 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/1029376 Reviewed-by: Ben Wagner <bungeman@google.com> Commit-Queue: Kaylee Lubick <kjlubick@google.com> | 11 个月前 | |
[skcapture] add initial plumbing for getting SkCapture data from clients This CL starts adding SkCaptureManager and SkCaptureCanvas references across Skia: * Providing an SkCapture factory for Surfaces to call on when capture is enabled. * Creating start and stop APIs for capture on the Context * Creating an external_client test file Change-Id: I42ccfaa71a3e3345d5661eb1d87e683e2c8d6346 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/1004621 Reviewed-by: Michael Ludwig <michaelludwig@google.com> Commit-Queue: Jorge Betancourt <jmbetancourt@google.com> | 11 个月前 | |
[graphite] Delete unused serialize/deserialize TextureInfo::Data funcs Change-Id: I56b86a80a312719abe212d8e03932578476306ba Reviewed-on: https://skia-review.googlesource.com/c/skia/+/1096757 Auto-Submit: Michael Ludwig <michaelludwig@google.com> Reviewed-by: Robert Phillips <robertphillips@google.com> Commit-Queue: Michael Ludwig <michaelludwig@google.com> | 7 个月前 | |
[graphite] Consolidate Utils/Types headers Public headers are consolidated to: foo/FooBackendContext.h - holding the backend context and context factory foo/FooGraphiteTypes.h - holding FooTextureInfo and factories for backend textures and semaphores Minor exceptions are for Metal, which requires the MtlGraphiteTypes header to be split into an ObjectiveC and C++-only header (this had been MtlGraphiteUtils.h but I renamed it to emphasize the language rationale). Vulkan also uses VulkanGraphiteContext.h since there is already a shared VulkanBackendContext.h. When Ganesh goes away, this can be consolidated. Internally, the variopus FooGraphiteTypesPriv, FooGraphiteUtilsPriv, and FooUtilsPriv are consolidated to just FooGraphiteUtils, with impl moving to FooGraphiteUtils.cpp as well. This is all just code moving and include updating. In follow up CLs, the context factory work will be moved out of FooGraphiteUtils.cpp and into FooContext.cpp. Everything related to FooTextureSpec will be removed entirely, instead just accessing a pointer to the public FooTextureInfo using TextureInfoPriv. Lastly, FooBackendTexture can likely undergo a similar consolidation. Bug: b/390473370 Change-Id: I242758467cfff3a926e1bafcf5a56c80af15020d Reviewed-on: https://skia-review.googlesource.com/c/skia/+/946258 Reviewed-by: Jim Van Verth <jvanverth@google.com> Commit-Queue: Michael Ludwig <michaelludwig@google.com> Reviewed-by: Greg Daniel <egdaniel@google.com> | 1 年前 | |
Reland "Reland "[graphite] Remove TextureInfoData and TextureSpec in TextureInfo"" This reverts commit cbc400181ddf1f73ad84532b1e868c62ffccf802. Reason for revert: fix ctor Original change's description: > Revert "Reland "[graphite] Remove TextureInfoData and TextureSpec in TextureInfo"" > > This reverts commit 2b944241704cbdee838fb73e4e9608dce0e44076. > > Reason for revert: compiler is thinking private template CTOR is > ambiguous. > > Original change's description: > > Reland "[graphite] Remove TextureInfoData and TextureSpec in TextureInfo" > > > > This reverts commit 4af764a24a7e2463c35fe681cc850a27baa3a697. > > > > Reason for revert: guard DawnTextureInfo change to class > > > > Original change's description: > > > Revert "[graphite] Remove TextureInfoData and TextureSpec in TextureInfo" > > > > > > This reverts commit 68be771f02f0c1b1a307fb6fafbcce003cb6d0fb. > > > > > > Reason for revert: must update forward declaration in chrome > > > > > > Original change's description: > > > > [graphite] Remove TextureInfoData and TextureSpec in TextureInfo > > > > > > > > The intermediates, TextureInfoData and its subclasses, and the backend > > > > structs [Foo]TextureSpec are removed. > > > > > > > > A private TextureInfo::Data virtual class is introduced that holds the > > > > sample count and mipmapped state, with the backend [Foo]TextureInfo > > > > classes extending. This exposes those fields and preserves their field > > > > layout before this CL, and also makes them equivalent to the removed > > > > TextureSpec structs. > > > > > > > > Since TextureInfo::Data is private to TextureInfo and can be friended, > > > > the virtual functionality that had been on TextureInfoData is declared > > > > up front to simplify the number of types that have to be implemented. > > > > > > > > In situations where there is a backend context (e.g. a Caps object) > > > > or calling a function from a backend-specific compilation unit, > > > > template traits are used instead of increasing the number of virtual > > > > functions. With a few more follow up CLs, the only virtual > > > > functionality on TextureInfo::Data will be related to equality and > > > > initialization/assignment operators. > > > > > > > > I opted to have the template functionality be part of the > > > > FooTextureInfo subclasses directly, instead of having a specialization > > > > for such as TextureInfoTraits<FooTextureInfo>. If it was that way, > > > > it would require all the callers/users of the templated functionality > > > > to include multiple headers and if the traits specialization wasn't > > > > included, the compiler error messages were pretty opaque. > > > > > > > > Lastly, this goes through all the backends and updates them to > > > > access their backend data directly using TextureInfoPriv::Get<T>, > > > > which just casts the underlying SkAnySubclass. > > > > > > > > > > > > Change-Id: I5f58146305175ca4da4d0feca5d8499cc850da35 > > > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/952676 > > > > Reviewed-by: Greg Daniel <egdaniel@google.com> > > > > Commit-Queue: Michael Ludwig <michaelludwig@google.com> > > > > > > Change-Id: I0c8a4962862af32bb720b56eb7f84e38ea901fe4 > > > No-Presubmit: true > > > No-Tree-Checks: true > > > No-Try: true > > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/954836 > > > Owners-Override: Kaylee Lubick <kjlubick@google.com> > > > Reviewed-by: Kaylee Lubick <kjlubick@google.com> > > > Auto-Submit: Michael Ludwig <michaelludwig@google.com> > > > > Change-Id: I25c933a8de920f382987b4ea5b81a312f14f6631 > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/955836 > > Reviewed-by: Robert Phillips <robertphillips@google.com> > > Reviewed-by: Kaylee Lubick <kjlubick@google.com> > > Commit-Queue: Michael Ludwig <michaelludwig@google.com> > > Change-Id: Ieeaccfab7b04e5031ebafebb7505f074942ab80a > No-Presubmit: true > No-Tree-Checks: true > No-Try: true > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/956296 > Auto-Submit: Michael Ludwig <michaelludwig@google.com> > Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> > Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Change-Id: Ieb21cd985d96f077087f69174f408e9b3a08f144 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/956297 Reviewed-by: Greg Daniel <egdaniel@google.com> Commit-Queue: Kaylee Lubick <kjlubick@google.com> Reviewed-by: Kaylee Lubick <kjlubick@google.com> Auto-Submit: Michael Ludwig <michaelludwig@google.com> | 1 年前 | |
[graphite] Consolidate Utils/Types headers Public headers are consolidated to: foo/FooBackendContext.h - holding the backend context and context factory foo/FooGraphiteTypes.h - holding FooTextureInfo and factories for backend textures and semaphores Minor exceptions are for Metal, which requires the MtlGraphiteTypes header to be split into an ObjectiveC and C++-only header (this had been MtlGraphiteUtils.h but I renamed it to emphasize the language rationale). Vulkan also uses VulkanGraphiteContext.h since there is already a shared VulkanBackendContext.h. When Ganesh goes away, this can be consolidated. Internally, the variopus FooGraphiteTypesPriv, FooGraphiteUtilsPriv, and FooUtilsPriv are consolidated to just FooGraphiteUtils, with impl moving to FooGraphiteUtils.cpp as well. This is all just code moving and include updating. In follow up CLs, the context factory work will be moved out of FooGraphiteUtils.cpp and into FooContext.cpp. Everything related to FooTextureSpec will be removed entirely, instead just accessing a pointer to the public FooTextureInfo using TextureInfoPriv. Lastly, FooBackendTexture can likely undergo a similar consolidation. Bug: b/390473370 Change-Id: I242758467cfff3a926e1bafcf5a56c80af15020d Reviewed-on: https://skia-review.googlesource.com/c/skia/+/946258 Reviewed-by: Jim Van Verth <jvanverth@google.com> Commit-Queue: Michael Ludwig <michaelludwig@google.com> Reviewed-by: Greg Daniel <egdaniel@google.com> | 1 年前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 11 个月前 | ||
| 11 个月前 | ||
| 7 个月前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 |