Pytorch 2.4.1, python 3.13, cudnn9

[regro-cf-autotick-bot]

This PR has been triggered in an effort to update python313.

Notes and instructions for merging this PR:

1. Please merge the PR only after the tests have passed.
2. Feel free to push to the bot's branch to update this PR if needed.

Please note that if you close this PR we presume that the feedstock has been rebuilt, so if you are going to perform the rebuild yourself don't close this PR until the your rebuild has been merged.

---

If this PR was opened in error or needs to be updated please add the bot-rerun label to this PR. The bot will close this PR and schedule another one. If you do not have permissions to add this label, you can use the phrase @conda-forge-admin, please rerun bot in a PR comment to have the conda-forge-admin add it for you.

_{This PR was created by the regro-cf-autotick-bot. The regro-cf-autotick-bot is a service to automatically track the dependency graph, migrate packages, and propose package version updates for conda-forge. Feel free to drop us a line if there are any issues! This PR was generated by - please use this URL for debugging.}

Also includes...

Fixes #264
Fixes #259

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[hmaarrfk]

if anybody could rebuild OSX that would be great!

Edit: nevermind, i forgot about
#256 and #259

[hmaarrfk]

cudnn 9 would make this package incompatible with cupy.
conda-forge/cupy-feedstock#286

Would it be appropriate to build out 2 versions, one for cudnn 9, the other for cudnn 8???

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found some lint.

Here's what I've got...

For recipe/meta.yaml:

• If python is a host requirement, it should be a run requirement.

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[hmaarrfk]

I thought that it i could get things to work by downgrading to 2.4.0 instead of 2.4.1.... but it still doesn't work

[Tobias-Fischer]

It looks like it's going ok so far? :)

[hmaarrfk]

No I ran it locally in docker on a linux-64 host and ran into the same problem as
#256

[jakirkham]

Am seeing the following error in this CI job (with attached log):

-- Using third party subdirectory Eigen.
CMake Error at /home/conda/feedstock_root/build_artifacts/libtorch_1728075572459/_build_env/share/cmake-3.30/Modules/FindPackageHandleStandardArgs.cmake:233 (message):
  Could NOT find Python (missing: Python_INCLUDE_DIRS Interpreter
  Development.Module NumPy) (found version "3.11.10")
Call Stack (most recent call first):
  /home/conda/feedstock_root/build_artifacts/libtorch_1728075572459/_build_env/share/cmake-3.30/Modules/FindPackageHandleStandardArgs.cmake:603 (_FPHSA_FAILURE_MESSAGE)
  /home/conda/feedstock_root/build_artifacts/libtorch_1728075572459/_build_env/share/cmake-3.30/Modules/FindPython.cmake:673 (find_package_handle_standard_args)
  cmake/Dependencies.cmake:857 (find_package)
  CMakeLists.txt:857 (include)

Edit: Another jobs fails with the same error

[hmaarrfk]

its probably because i hardcoded the numpy include dir and its frozen in.

[jakirkham]

What if we added a symlink to the NumPy headers using a Python version agnostic location and provided the latter as input?

[hmaarrfk]

you are welcome to try, i can back off.

[hmaarrfk]

actuallly my guess is that cmake is invoked through a new codepath i didn't capture in my patch.

[jakirkham]

I trust your judgment Mark. Please let me know if I'm just missing things

Edit: Fixed a typo

[hmaarrfk]

I’m too tired for today. Just giving ideas.

[hmaarrfk]

I think it is a difference between numpy 2.0 and 2.1.

Is there a reason to to build with 2.0 vs 2.1?

The reason i ask is that the migration pins to 2.0:
https://github.com/conda-forge/conda-forge-pinning-feedstock/blob/main/recipe/migrations/numpy2.yaml#L48

[h-vetinari]

Is there a reason to to build with 2.0 vs 2.1?

Just because we started out conservatively. You can use 2.1 without issue, or pin to 2 directly

[hmaarrfk]

i'm not so convinced about my 2.1 theory anymore. I had successful builds with Linux64 and aarch64 in the past...

[hmaarrfk]

@isuruf it seems the sed “fix” was at the source of the issue. Any other ideas on how to make sed better? Or should we just restart builds periodically.

[hmaarrfk]

Well thank you all!!!!!

[jakirkham]

Hooray! 🥳

Thanks everyone! 🙏