Make Linux (de)activate scripts standalone like on Windows

[leofang]

Supersede #68. Close #68.

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 (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

[conda-forge-linter]

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.

[leofang]

@conda-forge-admin, please rerender

[jakirkham]

Appreciate the effort here Leo, but am not sure this is going to work.

We rely on having some of these environment variables getting replaced while leaving others to be determined at runtime. If we go this route, it may require some patching of the scripts, using sed, and/or injecting the full paths into the scripts for prefix replacement.

If we find something that works here, would be good to do a fair bit of testing to make sure everything is still behaving as expected.

[leofang]

Hmmm I thought I did that, but maybe there're a few other variables that I missed. Please check the Windows scripts. The structure is identical now, so if we can do the variable injection on Windows, we should be able to do that on Linux too.

That said, I cannot continue today for some urgent tasks. Let me mark this as draft.

[jakirkham]

Gotcha. Yeah can take a look later

No worries. Thanks for the effort here. Agree this would be more readable/maintainable with this change

[leofang]

We rely on having some of these environment variables getting replaced while leaving others to be determined at runtime. If we go this route, it may require some patching of the scripts, using sed, and/or injecting the full paths into the scripts for prefix replacement.

Hi @jakirkham it appears to me that we only need to worry about PKG_VERSION (which was already correctly handled). Am I missing something?

[leofang]

@conda-forge-admin, please restart ci

[jakirkham]

@conda-forge-admin, please re-render

[github-actions]

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

I tried to rerender for you, but it looks like there was nothing to do.

This message was generated by GitHub actions workflow run https://github.com/conda-forge/nvcc-feedstock/actions/runs/2737010599.

[leofang]

@conda-forge-admin, please re-render

[jakirkham]

@galipremsagar @vyasr, could you please help test this when you have a chance?

Prebuilt Conda binaries can be found on CI (relevant docs). More context can also be found in issue ( conda-forge/conda-forge.github.io#1424 ).

[galipremsagar]

@galipremsagar @vyasr, could you please help test this when you have a chance?

Prebuilt Conda binaries can be found on CI (relevant docs). More context can also be found in issue ( conda-forge/conda-forge.github.io#1424 ).

Verified locally with the artifacts and works on my machine 👍

[jaimergp]

@leofang Do you think it's worth it to use set -euo pipefail in the (de)activation scripts? I am not sure if you considered some variables might be undefined (like CONDA_BUILD_SYSROOT) under certain circumstances.

[leofang]

@leofang Do you think it's worth it to use set -euo pipefail in the (de)activation scripts? I am not sure if you considered some variables might be undefined (like CONDA_BUILD_SYSROOT) under certain circumstances.

I'd like to leave this question to @jakirkham @galipremsagar. In #87 and other PRs we tried to make the (de)activation script work outside of conda-build. My only concern is if the (de)activation script fails conda may be left in an invalid state. I seem to remember seeing a similar issue in the past but can't recall where.

[leofang]

Another note is we currently have both CONDA_BUILD_SYSROOT and CONDA_BUILD used in the script. I aimed to do a simple migration in this PR so I didn't bother to touch them. Let me know if we need to unify the usage in this PR.

[vyasr]

@jakirkham @leofang @galipremsagar I was also able to verify that this worked for me locally.

[jakirkham]

@jaimergp, could you please take another look when you have a chance? 🙂

[leofang]

@conda-forge-admin, please re-render

[leofang]

Let's get this merged. Thanks everyone for reviewing & testing!