Add recipe for libmetatensor-torch

[Luthaf]

This is the continuation of conda-forge/metatensor-feedstock#2, now packaging the second native library. This one is pure C++, and links to libtorch.

Checklist

• Title of this PR is meaningful: e.g. "Adding my_nifty_package", not "updated meta.yaml".
• License file is packaged (see here for an example).
• Source is from official source.
• Package does not vendor other packages. (If a package uses the source of another package, they should be separate packages or the licenses of all packages need to be packaged).
• If static libraries are linked in, the license of the static library is packaged.
• Package does not ship static libraries. If static libraries are needed, follow CFEP-18.
• Build number is 0.
• A tarball (url) rather than a repo (e.g. git_url) is used in your recipe (see here for more details).
• GitHub users listed in the maintainer section have posted a comment confirming they are willing to be listed there.
• When in trouble, please check our knowledge base documentation before pinging a team.

[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 (recipes/libmetatensor-torch) and found some lint.

Here's what I've got...

For recipes/libmetatensor-torch:

• The following maintainers have not yet confirmed that they are willing to be listed here: PicoCentauri. Please ask them to comment on this PR if they are.

For recipes/libmetatensor-torch:

• This recipe is using a compiler, which now requires adding a build dependence on {{ stdlib("c") }} as well. For further details, please see META: {{ stdlib("c") }} migration conda-forge.github.io#2102.

[Luthaf]

Just to check, the version of libtorch is pinned across the ecosystem, right? And any bump in the libtorch version will trigger a rebuild of all feedstocks depending on it?

[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 (recipes/libmetatensor-torch) and found some lint.

Here's what I've got...

For recipes/libmetatensor-torch:

• The following maintainers have not yet confirmed that they are willing to be listed here: PicoCentauri. Please ask them to comment on this PR if they are.

[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 (recipes/libmetatensor-torch, recipes/python-metatensor) and found some lint.

Here's what I've got...

For recipes/libmetatensor-torch:

• The following maintainers have not yet confirmed that they are willing to be listed here: PicoCentauri. Please ask them to comment on this PR if they are.

For recipes/python-metatensor:

• The following maintainers have not yet confirmed that they are willing to be listed here: PicoCentauri. Please ask them to comment on this PR if they are.

[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 (recipes/libmetatensor-torch) and found some lint.

Here's what I've got...

For recipes/libmetatensor-torch:

• The following maintainers have not yet confirmed that they are willing to be listed here: PicoCentauri. Please ask them to comment on this PR if they are.

[Luthaf]

@conda-forge/help-c-cpp ready for review!

[xhochy]

Shared libraries also need a run_exports.

[Luthaf]

It's not very clear what run_exports does. Does it only add a run dependency to all packages that built against the library and had it as a host dependency? Or is there something else I need to be aware of here?

[carterbox]

https://docs.conda.io/projects/conda-build/en/latest/resources/define-metadata.html#export-runtime-requirements

run_exports export runtime requirements to downstream packages which dynamically link to your package. It's a statement which says "hey, if you link to my package, then this is the version that you will need at runtime."

[Luthaf]

Thanks! I've added run_exports to libmetatensor and removed it from the run requirements here.

I'd guess libtorch also has a run_exports, but is there a way to express "build against the cpu variant but run with any variant, cpu or cuda"? For now I do two separate host & run requirements.

[carterbox]

I'd guess libtorch also has a run_exports

Yes.

Looks like the run_exports for libtorch are variant agnostic, so if you build against any variant of libtorch you can get any variant at runtime. It looks like this package doesn't link against libtorch_cuda, so you don't need to worry about contraining the libtorch build variant.

[PicoCentauri]

Thanks for the work @Luthaf. Of course I am willing to help maintaining this recipe.

[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 (recipes/libmetatensor-torch) and found it was in an excellent condition.

[Luthaf]

@conda-forge/help-c-cpp ready for another round of review!

[carterbox]

The logs show that the build system is also looking for zlib? Is that a missing host dependency?

https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=935468&view=logs&j=6f142865-96c3-535c-b7ea-873d86b887bd&t=22b0682d-ab9e-55d7-9c79-49f3c3ba4823&l=808

[Luthaf]

The logs show that the build system is also looking for zlib? Is that a missing host dependency?

We are not trying to load ZLIB ourself. Digging around the CMake files, this is what seems to happen:

We load torch with find_package(Torch), which then does find_package(Caffe2)

https://github.com/pytorch/pytorch/blob/6c503f1dbbf9ef1bf99f19f0048c287f419df600/cmake/TorchConfig.cmake.in#L68

Which then does find_package(Protobuf):

https://github.com/pytorch/pytorch/blob/4ed93d6e0c5deb543ba5a3bd103728f00d39b1a6/cmake/Caffe2Config.cmake.in#L48
https://github.com/pytorch/pytorch/blob/6c503f1dbbf9ef1bf99f19f0048c287f419df600/cmake/public/protobuf.cmake#L4

which finally tries to find ZLIB:

https://github.com/protocolbuffers/protobuf/blob/2f3242c576504459621fd7d78bbccf39bfaa49c5/cmake/protobuf-config.cmake.in#L5
https://github.com/protocolbuffers/protobuf/blob/2f3242c576504459621fd7d78bbccf39bfaa49c5/CMakeLists.txt#L172