Do not use _GNU_SOURCE gratuitously.

[przemoc]

What is needed to build whisper.cpp and examples is availability of stuff defined in The Open Group Base Specifications Issue 6 (https://pubs.opengroup.org/onlinepubs/009695399/) known also as Single Unix Specification v3 (SUSv3) or POSIX.1-2001 + XSI extensions.

There is no need to penalize musl libc which simply follows standards.

Not having feature test macros in source code gives greater flexibility to those wanting to reuse it in 3rd party app, as they can build it with minimal FTM (_XOPEN_SOURCE=600) or other FTM depending on their needs.

It builds without issues in Alpine (musl libc), Ubuntu (glibc), MSYS2.

[ggerganov]

The macOS-latest action seems to fail with this change - SDL2 is not happy for some reason

[przemoc]

Thanks for running CI checks. I do not have access to any macOS platform, so debugging it may be a bit challenging.

I whisper.cpp build info: 
I UNAME_S:  Darwin
I UNAME_P:  i386
I UNAME_M:  x86_64
I CFLAGS:   -I.              -O3 -DNDEBUG -std=c11   -fPIC -D_XOPEN_SOURCE=600 -pthread -mf16c -mavx -DGGML_USE_ACCELERATE
I CXXFLAGS: -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -D_XOPEN_SOURCE=600 -pthread
I LDFLAGS:   -framework Accelerate
I CC:       Apple clang version 14.0.0 (clang-1400.0.29.202)
I CXX:      Apple clang version 14.0.0 (clang-1400.0.29.202)

c++ -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -D_XOPEN_SOURCE=600 -pthread examples/stream/stream.cpp examples/common.cpp examples/common-ggml.cpp examples/common-sdl.cpp ggml.o whisper.o -o stream `sdl2-config --cflags --libs`  -framework Accelerate
In file included from examples/stream/stream.cpp:7:
In file included from ./examples/common-sdl.h:3:
In file included from /usr/local/include/SDL2/SDL.h:32:
In file included from /usr/local/include/SDL2/SDL_main.h:25:
/usr/local/include/SDL2/SDL_stdinc.h:532:5: error: use of undeclared identifier 'memset_pattern4'
    memset_pattern4(dst, &val, dwords * 4);
    ^
1 error generated.

[przemoc]

Looks like my attempt at fixing macOS issue was successful, others checks will be most likely successful too, but let's wait for them to finish.

[ggerganov]

I synced the latest ggml from llama.cpp and the new NUMA-related changes fail to compile:

https://github.com/ggerganov/whisper.cpp/actions/runs/5438047714/jobs/9889005968

Temporary reverted the change until we resolve the build issues.
I think we should first fix ggml-org/llama.cpp#2035 and then I will propagate the changes back to whisper.cpp

[przemoc]

Sorry, @ggerganov, it turned out I didn't have time this weekend to play with llama.cpp and cmake. :(
Adding selected lines in 1st commit from ggml-org/llama.cpp#2035 would most likely solve failing build in whisper.cpp, but of course only for Makefile-driven builds.

BTW What's cmake usage state in whisper.cpp? I don't think CI was doing any cmake-driven build last time I've seen those PR actions. Is it expected to be working? If yes, then I guess I may have broken it with my FTM sanitization patches that weren't touching CMakeLists.txt.

[ggerganov]

The main purpose for having CMake build system is to support Windows.

I think the CI wasn't failing before because the NUMA changes were not present.
Yesterday I synced them from the llama.cpp repo and they started causing problems.

Don't worry - whenever you have the time. It's not a big problem

[przemoc]

I looked at llama.cpp a bit on Sunday and went with Howard's approach for CMake changes to simply follow recent FTM improvements from Makefile - seemed like viable way way of doing that. I had to fix other things first, though, to make it build in MSYS2.

@ggerganov, when you'll have a spare minute, please trigger PR checks in ggml-org/llama.cpp#2035 and the other one mentioned there that should be merged first, so that I could see if there are still any issues that I should resolve.

And when that will be clean, I can create analogous remove-gnu-source-ftm for whisper.cpp. Not sure if GitHub allows merging second time from same PR (considering it was reverted), as I would prefer to have updated rebased version here and not create yet another PR.

[przemoc]

Thanks for triggering PR checks in llama.cpp. Stuff seem to be building fine there. I ended up creating new PR in whisper.cpp, #1129, to avoid potential issues with GitHub merging UX.