mvapich-plus: v4.0 hash changed (?)

[kftsehk]

Found the sha256 changed, as is downloaded from the webpage as well.

Version is still named "4.0"

[alalazo]

I see the same. @natshineman @MatthewLieber do you know why?

[MatthewLieber]

Hi it looks like you are confusing our 2 packages. That is the sha for the mvapich package not the mvapich-plus package. There is no publicly available tar for mvapich-plus as it is a closed source release. We have buildcaches available on our mirror http://mvapich.cse.ohio-state.edu/download/mvapich/spack-mirror/mvapich-plus for users to install and if you would like a version not in the mirror please let us know and we can generate a new one for you.

[alalazo]

@MatthewLieber Thanks for the explanation. The fact is that, if you do:

$ spack checksum mvapich-plus 4.0
==> Found 1 version of mvapich-plus
==> Fetching https://mvapich.cse.ohio-state.edu/download/mvapich/mv2/mvapich-4.0.tar.gz

    version("4.0", sha256="c532f7bdd5cca71f78c12e0885c492f6e276e283711806c84d0b0f80bb3e3b74")

so this is indeed a bit confusing. How is this package usually fetched and built by users?

[MatthewLieber]

They should install the buildcache from our mirror if they want the closed source version.

[alalazo]

@MatthewLieber So am I understanding correctly that this is the recipe you use to create your closed source buildcache? And the sha is the sha256 of the closed source tar?

I'm just trying to think how we can prevent people from being able to checksum this package, since its url points to regular mvapich, and raise false flags on checksum mismatch.

[MatthewLieber]

Yes that is correct. Can we fix this by changing the url to be one that does not exist or will that break things further?

[kftsehk]

I think we can set it to private/local source tarball required. For example, I find this example from matlab that requires local (propietary) installer to install by setting

    extendable = True
    manual_download = True
    
    def url_for_version(self, version):
        if version >= Version("R2024"):
            return "file://{0}/matlab_{1}_Linux.zip".format(os.getcwd(), version)
        return "file://{0}/matlab_{1}_glnxa64.zip".format(os.getcwd(), version)

user would then require to manual download the sha256-matching source, place it in (by-default) ./<some-standard-name>.<ext>

This approach would allow administrators on clusters which has access to the source to make a local source mirror.

[kftsehk]

By the way, I also find confusing that we have 4 mvapich packages, mvapich2 mvapich2-x mvapichX and mvapich-plus.

I believe they are variants and versions of mvapich, as I understand

• X is a more advance configuration capable of providing various GDR/NCCL/RoCM/RoCE support
• Plus is even more development edge, which may provcide specific hardware support
• There are also versions of 2.x, 3.x and recently 4.0, which implements different interface of MPICH

[haampie]

Hi @kftsehk, this package-related pull request can be migrated from spack/spack to spack/spack-packages using the migrate-pkg-prs utility.

We encourage authors to run the migration script themselves to preserve author attribution.

You can migrate all your open pull requests at once, following the steps in the documentation.

---

This comment was posted automatically.

[github-actions[bot]]

This pull request has been automatically marked as stale because it has not had any activity in the last 6 months. It will be closed in 30 days if there is no further activity.

If the pull request is waiting for a reply from reviewers, feel free to ping them as a reminder. If it is waiting and has no assigned reviewer, feel free to ping @spack/spack-releasers or simply leave a comment saying this should not be marked stale. This will reset the pull request's stale state.

To get more eyes on your pull request, you can post a link in the #pull-requests channel of the Spack Slack. Thank you for your contributions!