Skip to content

Patch for building [email protected], [email protected], and [email protected]:2.61 with Intel oneAPI 2025.2; add [email protected]; intel-oneapi-runtime: remove when constraint for providing fortran-rt and libifcore#4

Merged
climbfuji merged 2 commits intoJCSDA:spack-stack-devfrom
climbfuji:bugfix/oneapi2025p2_nasa_packages
Oct 3, 2025

Conversation

@climbfuji
Copy link
Copy Markdown
Collaborator

@climbfuji climbfuji commented Sep 30, 2025

…neapi_runtime/package.py to be able to use external intel-oneapi-runtime
@climbfuji climbfuji self-assigned this Oct 2, 2025
@climbfuji climbfuji changed the title WIP Bugfix/oneapi2025p2 NASA packages + external intel-onapi-runtime Patch for building [email protected], [email protected], and [email protected]:2.61 with Intel oneAPI 2025.2; add [email protected]; intel-oneapi-runtime: remove when constraint for providing fortran-rt and libifcore Oct 2, 2025
@climbfuji climbfuji marked this pull request as ready for review October 2, 2025 13:10
@climbfuji climbfuji moved this to In Progress in spack-stack-2.0.x (2025 Q4) Oct 2, 2025
@mathomp4
Copy link
Copy Markdown
Collaborator

mathomp4 commented Oct 2, 2025

@climbfuji You might want to look at some of the changes in spack#1782 as well. For example for yafyaml I had to add:

    # This is needed because for ifx 2025.2, we need to use cpp from GNU as fpp from oneapi
    # is broken. To pull that in, we need to say yafyaml depends on C, even though it really
    # doesn't.
    depends_on("c", when="^[email protected]", type="build")
    depends_on("gcc", when="^[email protected]", type="build")

because while yafyaml v1.6.0 has the fpp fix built into its CMake, it still needs the "external" cpp. I based this on what you did in spack#1592 for MAPL.

@climbfuji
Copy link
Copy Markdown
Collaborator Author

@climbfuji You might want to look at some of the changes in spack#1782 as well. For example for yafyaml I had to add:

    # This is needed because for ifx 2025.2, we need to use cpp from GNU as fpp from oneapi
    # is broken. To pull that in, we need to say yafyaml depends on C, even though it really
    # doesn't.
    depends_on("c", when="^[email protected]", type="build")
    depends_on("gcc", when="^[email protected]", type="build")

because while yafyaml v1.6.0 has the fpp fix built into its CMake, it still needs the "external" cpp. I based this on what you did in spack#1592 for MAPL.

If my upstream spack PR is no longer valid, then perhaps we should close it and use your PR instead? I can update the packages here from your upstream PR.

@climbfuji
Copy link
Copy Markdown
Collaborator Author

@climbfuji You might want to look at some of the changes in spack#1782 as well. For example for yafyaml I had to add:

    # This is needed because for ifx 2025.2, we need to use cpp from GNU as fpp from oneapi
    # is broken. To pull that in, we need to say yafyaml depends on C, even though it really
    # doesn't.
    depends_on("c", when="^[email protected]", type="build")
    depends_on("gcc", when="^[email protected]", type="build")

because while yafyaml v1.6.0 has the fpp fix built into its CMake, it still needs the "external" cpp. I based this on what you did in spack#1592 for MAPL.

If my upstream spack PR is no longer valid, then perhaps we should close it and use your PR instead? I can update the packages here from your upstream PR.

@mathomp4 Another option is to proceed with this PR as is for spack-stack-dev. We don't have yafyaml 1.6.0 anyway.

Then, we close my upstream PR and let your GFE packages update PR handle all the packages. Once your PR gets merged, we can update all the GFE packages by cherry-picking the squashed merge and (via conflict resolution) overwrite my changes in this PR.

But I can also do this now from your PR by cherry-picking all your commits (I can't pull your branch, because spack-stack-dev is currently behind develop).

What's your preference?

@mathomp4
Copy link
Copy Markdown
Collaborator

mathomp4 commented Oct 3, 2025

@climbfuji Oh yeah. I guess if you don't have yafyaml 1.6.0 it's sort of moot. So, yeah, stick with this and then when spack takes up the PRs in their repo, we can update here.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

No open projects

Development

Successfully merging this pull request may close these issues.

2 participants