hdf5: conflicts zlib-ng+new_strategies

[skosukhin]

The package should not be built against zlib-ng+new_strategies (see HDFGroup/hdf5#4147). As far as I understand, the fix is backported and will be available in @1.14.4.

[spackbot-app]

@byrnHDF can you review this PR?

This PR modifies the following package(s), for which you are listed as a maintainer:

• hdf5

[skosukhin]

I would prefer the following instead:

conflicts("^zlib-ng+new_strategies", when="@:1.14.3,develop-1.8:develop-1.12")

It concretizes fine for spack spec hdf5, i.e. it gives hdf5@1.14.3 ^zlib-ng~new_strategies.
However, spack spec netcdf-c prefers the develop version of hdf5: netcdf-c ^hdf5@develop-1.15 ^zlib-ng+new_strategies, which is technically correct but undesirable.

@alalazo do you think there is anything reasonable that can be done so that the develop versions are deprioritized (even more)?

[alalazo]

@spackbot run pipeline

[spackbot-app]

I've started that pipeline for you!

[alalazo]

@spackbot run pipeline

[spackbot-app]

I've started that pipeline for you!

[alalazo]

@johnwparent @kwryankrattiger Fine with you to force merge this PR? The Windows failure on Gitlab is unrelated. Or is there any way to make this pass Gitlab CI?

[kwryankrattiger]

Since this is changing HDF5 and the config for zlib-ng we are using, I would like to see it build and run tests before merge since this is a core library, especially for the windows pipeline.

The issue is related to PR build cache. The fix should be to simply close and reopen the PR and let CI run.

[johnwparent]

Agreed, would also like to see any validation that this passes on Windows.

The issue is related to PR build cache. The fix should be to simply close and reopen the PR and let CI run.

That was not sufficient in my experience, I had to create a new PR, as closing cleared the build cache, but generate was continuing to generate the same hashes, so every stack started missing.

[kwryankrattiger]

@spackbot run pipeline

[spackbot-app]

I'm sorry, gitlab does not have your latest revision yet, I can't run that pipeline for you right now.

One likely possibility is that your PR pipeline has been temporarily deferred, in which case, it is awaiting a develop pipeline, and will be run when that finishes.

Please check the gitlab commit status message to see if more information is available.

Details

Unexpected response from gitlab: {'message': '404 Commit Not Found'}

[kwryankrattiger]

You shouldn't need a new PR, as the spackbot reply shows the branch is gone on gitlab as well, so sync will make a new merge commit with the latest develop and push that. The close PR action makes the re-opened PR look the same as a new PR with the same commit. If the robot doesn't pick up the event then this wouldn't work, which may happen sometimes.

[kwryankrattiger]

Once this finishes we will need to re-run again :/ not sure why but gitlab is dropping trigger pipelines which is a new one for me.

[kwryankrattiger]

@spackbot run pipeline

[spackbot-app]

I've started that pipeline for you!

[kwryankrattiger]

It looks like the windows are past the zlib-ng/HDF5 issues.

The other pipelines are also looking okay, but will probably take a while since we cleared all of the new stuff from the PR cache.

Let me know if this is still stuck, otherwise LGTM!