Improve CMake support

[cNoNim]

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy
• Ensured the license file is being packaged.

Closes #7

[cNoNim]

@conda-forge-admin, please ping conda-forge/core

Looks like I broke the noarch.
I read https://conda-forge.org/docs/maintainer/knowledge_base.html#noarch-builds

And actually, we have two packages.

1. glad2 - noarch: python
2. glad2-cmake requires OS-specific build scripts to place the cmake package in different prefixes for Windows and Unix.

How can it be fixed?

I can try putting cmake-package in both locations, but not fully understand how i can do that without scripts.

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-webservice.

I was asked to ping @conda-forge/core and so here I am doing that.

[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) and found it was in an excellent condition.

[cNoNim]

@traversaro

The last iteration seems to have worked.
Going into production?

[cNoNim]

@conda-forge-admin, please rerender

[traversaro]

@traversaro

The last iteration seems to have worked. Going into production?

Interesting! To be honest I am afraid that in this configuration the pin_subpackage('glad2', exact=True) will fail once a new glad2 version will be released, as for example for glad2-cmake==3.0.0 there will be no glad2==3.0.0 available on non-Linux platforms. However, you can handle that once a new glad2 version is released. Anyhow, now I understand a bit better the problem, thanks! I have a similar situation in https://github.com/conda-forge/gz-msgs-feedstock/blob/main/recipe/meta.yaml#L66, but to be honest I never realized the problem: indeed multiple noarch package with the same version and build numbers are uploaded, and at the moment I never had problems with that, but I understand that the situation may be confusing. Perhaps a possible solution to avoid uploading multiple noarch packages is to manually force the build string to be a given unique value, so that even if each operating system creates its own noarch package, then only a single one will be uploaded, and all other will be skipped.

[cNoNim]

Interesting! To be honest I am afraid that in this configuration the pin_subpackage('glad2', exact=True) will fail once a new glad2 version will be released, as for example for glad2-cmake==3.0.0 there will be no glad2==3.0.0 available on non-Linux platforms.

Ouch... So because I made pin_subpackage('glad2', exact=True), the dependency is not allowed on platforms other than Linux.
This is my mistake, I think it needs to be fixed even now.
I'll try to test on a clean environment now.

[cNoNim]

Seems to have worked.

PS E:\Projects\test-glad2-cmake> micromamba create -p ./.mamba
Empty environment created at prefix: .mamba
PS E:\Projects\test-glad2-cmake> micromamba activate .\.mamba\
(.mamba)PS E:\Projects\test-glad2-cmake> micromamba install glad2-cmake
nvidia/noarch                                                 No change
pytorch/win-64                                                No change
pytorch/noarch                                                No change
nvidia/win-64                                                 No change
pkgs/msys2/win-64                                             No change
pkgs/msys2/noarch                                             No change
pkgs/r/noarch                                                 No change
pkgs/main/noarch                                              No change
pkgs/r/win-64                                                 No change
pkgs/main/win-64                                              No change
conda-forge/win-64                                  20.7MB @  29.9MB/s  0.7s
conda-forge/noarch                                  13.4MB @   9.7MB/s  1.1s

Transaction

  Prefix: E:\Projects\test-glad2-cmake\.mamba

  Updating specs:

   - glad2-cmake


  Package                 Version  Build               Channel           Size
-------------------------------------------------------------------------------
  Install:
-------------------------------------------------------------------------------

  + libexpat                2.5.0  h63175ca_1          conda-forge     Cached
  + ucrt             10.0.22621.0  h57928b3_0          conda-forge     Cached
  + ca-certificates    2023.11.17  h56e8100_0          conda-forge     Cached
  + python_abi               3.12  4_cp312             conda-forge     Cached
  + vc14_runtime      14.38.33130  h82b7239_18         conda-forge     Cached
  + vs2015_runtime    14.38.33130  hcb4865c_18         conda-forge     Cached
  + vc                       14.3  hcf57466_18         conda-forge     Cached
  + tk                     8.6.13  h5226925_1          conda-forge     Cached
  + openssl                 3.2.1  hcfcfb64_0          conda-forge     Cached
  + libzlib                1.2.13  hcfcfb64_5          conda-forge     Cached
  + libffi                  3.4.2  h8ffe710_5          conda-forge     Cached
  + bzip2                   1.0.8  hcfcfb64_5          conda-forge     Cached
  + libsqlite              3.44.2  hcfcfb64_0          conda-forge     Cached
  + xz                      5.2.6  h8d14728_0          conda-forge     Cached
  + tzdata                  2023d  h0c530f3_0          conda-forge     Cached
  + python                 3.12.1  h2628c8c_1_cpython  conda-forge     Cached
  + markupsafe              2.1.4  py312he70551f_0     conda-forge     Cached
  + jinja2                  3.1.3  pyhd8ed1ab_0        conda-forge     Cached
  + glad2                   2.0.4  pyha770c72_1        conda-forge      396kB
  + glad2-cmake             2.0.4  h57928b3_1          conda-forge       11kB

  Summary:

  Install: 20 packages

  Total download: 407kB

-------------------------------------------------------------------------------

[traversaro]

Yes, it works fine now. But as soon as a new version of glad is released, the build of glad2-cmake will fail on platforms different from linux64.