Hello, On 2023-03-20 07:01, M. Zhou wrote:
On Wed, 2023-03-15 at 15:50 +0200, Andrius Merkys wrote:I see. Nevertheless, thanks a lot for your guidance up to this point.I fixed the install target by updating the cmake patches. I think I need to forward these patches to the upstream later. Now I'm stuck here, which is exactly the previous problem that you have encountered: https://lists.debian.org/debian-mentors/2023/03/msg00039.html I did not realize it was a dependency resolving issue in our dev tools back then.
I saw your message on debian-mentors@. Maybe a quick fix would be to skip dh_shlibdeps step entirely for now.
After sorting everything out, I plan to gradually get the remaining reverse dependencies for the cuda variant into experimental. The code on the pytorch-cuda branch will be merged into master. I'll create a new `cuda` branch for the real packaging that will be uploaded to the archive (I need to double check the licenses before doing this).
This sounds good.
BTW, PyTorch 2.0 is available -- along with a batch of new dependency libraries that do not seem easy to handle. I'll stick to PyTorch 1.X for a while untill the start of python 3.12 transition.
OK.I recently started getting FTBFS for tensorpipe due to nonexistent cuda-gcc and cuda-g++. I see they come from nvidia-cuda-toolkit-gcc which is commented out in [1]. So 'cuda' branch no longer builds the cuda version, is this intentional?
[1] https://salsa.debian.org/deeplearning-team/tensorpipe/-/commit/bb101360436c79279510822497ee25f563702801
Best, Andrius