Installing python-pytorch-cuda requires gcc13 but gcc is already installed

Hello,

I’m trying to install the python-pytorch-cuda package using pacman but it depends on the extra/gcc13 and extra/gcc13-libs packages. My system already has gcc installed (and it’s actually the version 13.2.1 20240417) so I get multiple xxxx file exists in filesystem (owned by gcc) errors during the pytorch installation, the last one being:

gcc13: /usr/share/gcc-13.2.1/python/libstdcxx/v6/xmethods.py exists in filesystem (owned by gcc)

I tried uninstalling gcc but it breaks various packages:

gcc
├─base-devel
├─clang
│ ├─openshadinglanguage
│ │ ├─blender
│ │ └─usd
│ │   └─blender
│ └─shiboken6
│   └─pyside6
│     └─usd
└─dkms
  ├─nvidia-dkms
  └─nvidia-inst

Do you have any suggestion? Is there a way to tell pacman to use an already installed package? Could I force the gcc removal and then substitute it with gcc13?

Uninstall gcc13 and then it should reinstall with the python package, or:

sudo pacman -S  python-pytorch-cuda --overwrite='*'