AaronBallman wrote: > > We should validate that it can round trip serialization correctly, > > Can you clarify what you mean by this. I test serialization in > `AST/spirv-event-type.cpp`, similar to what is done here - #94830
Thanks! I was thinking of modules more than PCH, but CC @ChuanqiXu9 and @Bigcheese as serialization code owners; whatever they think is reasonable is fine by me. > > test mangling behavior (oops, probably need to update the demangler as > > well), etc. > > Mangling has been tested in `CodeGenCXX/spirv-event-type.cpp`. I am not sure > how to write a test for demangling. I could not find tests for opaque > builtins I based this patch on. FWIW I tested locally and it works. We don't > need to update the demangler since we use u mangling and this is already > handled by the demangler. > > ``` > $ ./build/bin/llvm-cxxfilt _Z12consumeEventu15__spirv_event_t > consumeEvent(__spirv_event_t) > > $ ./build/bin/llvm-cxxfilt _Z4funcIu15__spirv_event_tEvT_ > void func<__spirv_event_t>(__spirv_event_t) > ``` Ah, that's good! The demangler is actually an LLVM thing rather than a Clang thing: https://github.com/llvm/llvm-project/tree/main/llvm/tools/llvm-cxxfilt but if this is handled via generic means because we're using an extension mangling, that seems fine. https://github.com/llvm/llvm-project/pull/207077 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
