friendlyanon

Results 97 comments of friendlyanon

> The intention of this PR is not to support Windows. It is just one building block to get there. Sure, but why do extra work now that later has...

When you say "I don't want to make a switch from Make to CMake immediately" are you also considering users in this? As in, you don't want to make it...

> No such option exists for `cmake` as far as I know. `cmake --build` **requires** a path argument: https://cmake.org/cmake/help/latest/manual/cmake.1.html#build-a-project

@rui314 The below script successfully builds mold and its dependencies (with the above fix): ```sh #!/bin/bash set -euo pipefail git clone [email protected]:microsoft/mimalloc.git ( cd mimalloc cmake -GNinja -DBUILD_SHARED_LIBS=1 -DCMAKE_BUILD_TYPE=Release -B...

I built mimalloc and oneTBB from source as you can see, so I got the CMake package from them proper. However, I have sent an email to @alexmyczko (the Debian...

Oh I see, there is something fishy going on with the Debian packages. [libmimalloc2.0](https://packages.debian.org/bookworm/amd64/libmimalloc2.0/filelist) contains the CMake package, but that is missing from [libmimalloc-dev](https://packages.debian.org/bookworm/amd64/libmimalloc-dev/filelist). I think this is an actual...

I checked the -dev package and what mimalloc itself builds, and I'm not sure how the build rules on salsa achieve what goes into the .deb. Expected output from mimalloc...

While the paths are correct now, your bugreport was requesting mimalloc headers to be directly in the system includes of compilers. mimalloc upstream build file allows this without patches, so...

Hmm, I thought about the pkgconfig code and it's actually not quite right as it is now either. The problem with generating a pkgconfig file via CMake is that it...

You are correct. The project doesn't use the normal test mechanisms for whatever reason. https://github.com/hanickadot/compile-time-regular-expressions/blob/b51d3290865615487cbef75c7c9dba36b9e70c6a/tests/CMakeLists.txt#L1-L10 Note the lack of `add_test` calls. I think that could be added to this PR...