It's failing with CMAKE_C_COMPILER_AR-NOTFOUND
on the Chaotic build server too.
I don't think turning off LTO or using GCC are viable solutions. I vote for the -DCMAKE_AR=/usr/bin/llvm-ar
option. (And maybe llvm-ranlib too?)
Git Clone URL: | https://aur.archlinux.org/pcsx2-git.git (read-only, click to copy) |
---|---|
Package Base: | pcsx2-git |
Description: | PlayStation 2 emulator |
Upstream URL: | https://github.com/PCSX2/pcsx2 |
Licenses: | GPL-3.0-or-later |
Conflicts: | pcsx2 |
Provides: | pcsx2 |
Submitter: | alucryd |
Maintainer: | weirdbeard (xiota) |
Last Packager: | weirdbeard |
Votes: | 131 |
Popularity: | 0.35 |
First Submitted: | 2014-03-26 14:17 (UTC) |
Last Updated: | 2025-03-25 20:47 (UTC) |
« First ‹ Previous 1 .. 16 17 18 19 20 21 22 23 24 25 26 .. 70 Next › Last »
It's failing with CMAKE_C_COMPILER_AR-NOTFOUND
on the Chaotic build server too.
I don't think turning off LTO or using GCC are viable solutions. I vote for the -DCMAKE_AR=/usr/bin/llvm-ar
option. (And maybe llvm-ranlib too?)
Can I get a count of how many are experiencing this AR issue? Reason I'm asking is because this is a very ... Inconsistent issue meaning it doesn't happen on every machine and it doesn't happen every time. I can make it happen to me if I drop lld but it'd work on another machine vice versa. The only solid fix to this is either turning off LTO, or using GCC. Stenznek is very insistent that every thing be clang LTO which is why we're experiencing this.
It works for me, and KamFrento on the discord. I can switch to LLVM AR but I'm not certain that's the real fix as much as a work around for some given how sporadic and random it happens. I can't emphasize that enough as much of an unfortunate situation as that is.
For me build was failing with CMAKE_C_COMPILER_AR-NOTFOUND
, CMAKE_CXX_COMPILER_AR-NOTFOUND
and CMAKE_C_COMPILER_RANLIB-NOTFOUND
errors. I was able to make it work by addition of llvm
to makedepends
and setting -DCMAKE_AR=/usr/bin/llvm-ar
and -DCMAKE_RANLIB=/usr/bin/llvm-ranlib
cmake options.
Alright, since despite it doing so on my machine it's then clear that it's not on others. I've moved the exports to cmake options. Should resolve that issue
It's not using clang on my machine. Tried in a chroot and still is using gcc.
It definitely is. Earlier it was crashing on some old clang specific code in gtest
I still don't think it's using clang.
-- The C compiler identification is GNU 12.2.0
-- The CXX compiler identification is GNU 12.2.0
Maybe we need to pass options to cmake like this?
-DCMAKE_CXX_COMPILER=clang++ \
-DCMAKE_C_COMPILER=clang \
-DCMAKE_C_FLAGS="$CFLAGS -flto=thin" \
-DCMAKE_CXX_FLAGS="$CXXFLAGS -flto=thin" \
Or maybe export those envvars in the build function?
Try it again. I was waiting for a PR that would fix a few things I was having trouble with / it'll actually build with clang now if it wasn't before.
Its failing to compile for me. cc1plus: all warnings being treated as errors [626/694] Building CXX object pcsx2/CMakeFiles/GS-avx2.dir/GS/Renderers/SW/GSSetupPrimCodeGenerator.all.cpp.o ninja: build stopped: subcommand failed.
Oops, yeah sorry. Thanks for catching that mistake. Been going back and forth on keeping with the changes to PCSX2's build system and missed that. Otherwise yeah, clang + LTO's a bit faster then GCC which is why it's set
Pinned Comments
weirdbeard commented on 2024-08-17 03:40 (UTC)
https://github.com/PCSX2/pcsx2/pull/11632
This package now enables Cmake Package mode proper. PCSX2 will here on, be installed in the package standard folders /usr/bin, /usr/share, /usr/lib. Following the XDG standard pcsx2's config files remain in .config/PCSX2
In order to ensure a proper and clean upgrade. Uninstall this package COMPLETELY and clear cache before reinstalling.