After update in PR #169281 (commit f394cc6c5) it won't build:
https://hydra.nixos.org/build/174323623
It's still unreliable, as even the i686 builds that succeeded on Hydra
won't build for me locally, shooting over the 32-bit memory limit.
But I trust this will unblock the nixos-unstable channel on Hydra.
libpulseaudio doesn't currently build on darwin after a recent
upgrade (#160097). When looking closely at the darwin build, it
appeared that it wasn't being used at all.
❯ nix log $(nix-build --no-out-link channel:nixpkgs-unstable --argstr system x86_64-darwin -A qt512.qtmultimedia) | grep PulseAudio
Checking for PulseAudio >= 0.9.10... no
PulseAudio ............................. no
❯ nix log $(nix-build --no-out-link channel:nixpkgs-unstable --argstr system x86_64-darwin -A qt514.qtmultimedia) | grep PulseAudio
Checking for PulseAudio >= 0.9.10... no
PulseAudio ............................. no
❯ nix log $(nix-build --no-out-link channel:nixpkgs-unstable --argstr system x86_64-darwin -A qt515.qtmultimedia) | grep PulseAudio
Checking for PulseAudio >= 0.9.10... no
PulseAudio ............................. no
* cudaPackages.cudnn: migrate to redist cudaPackages
* cudaPackages.cudnn: fix missing zlib in rpath
* cudaPackages.cudnn: remove the propagated cudatoolkit
* cudaPackages.cudnn: use autoPatchelfHook
instead of custom find ... -exec ldd | grep routine
mark libcudnn_cnn_infer.so as needed for libcudnn.so on cudnn>=8.0.5
- a hint for autoPatchelf, as an alternative to manually adding $ORIGIN
* cudaPackages.cudnn: use automatic hooks for rpath
as a more common way to use addOpenGLRunpath and autoPatchelf with cudaPackages
* cudaPackages.cudnn: consume individual cuda packages
...since cudnn is part of the cuda package set
- introduces the scary useCudatoolkitRunfile function argument
to discourage usage of the runfile-based cudatoolkit
- instead of the rather hidden useRedist term in let ... in
- repeats cudatoolkit_root pattern after cuda_joined in pytorch &c
(the "toolchain view")
- redist packages are marked optional to support cuda<11.4 where the
attributes for redist packages do not exist
* cudaPackages.cudnn: update to pname+version
Co-authored-by: Sandro <sandro.jaeckel@gmail.com>
Co-authored-by: Sandro <sandro.jaeckel@gmail.com>
Apparently on x86_64-darwin there are also C files compiled,
and that doesn't go well with -std=c++*
https://hydra.nixos.org/build/174111870
Fortunately the flag only seems needed with gcc 11 so far.
