support for FFTW

[MarDiehl]

@conda-forge-admin, please rerender

Checklist

• [x] Used a personal fork of the feedstock to propose changes
• [x] Bumped the build number (if the version is unchanged)
• [x] Reset the build number to 0 (if the version changed)
• [x] Re-rendered with the latest conda-smithy (Use the phrase code>@<space/conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• [x] Ensured the license file is being packaged. (nothing has changed)

[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.

[dalcinl]

@MarDiehl LGTM. Please review a few build logs to double check. Ping me back once ready and I'll merge.

[dalcinl]

@MarDiehl Perhaps you need to specify a variant string to use MPI-enabled FFT builds?

[dalcinl]

Or perhaps via conda_build_config.yml, as done with hdf5 for the MPI variant.

[MarDiehl]

Or perhaps via conda_build_config.yml, as done with hdf5 for the MPI variant.

I think this is the correct way. I'll give it a try

[dalcinl]

@isuruf Could you please tell us what's the proper way to make this recipe build and depend on the MPI-enabled FFTW package?

[MarDiehl]

Looks ok now, the only failed test is the result of a timeout. Anything do do from my side?

[dalcinl]

@MarDiehl Let's start fresh, revert all of your changes to the meta.yaml and conda_build_config.yaml, and specify the fftw package the following way

requirements:
  host:
    - fftw * *mpi_{{ mpi }}*
  run:
    - fftw * *mpi_{{ mpi }}*

[MarDiehl]

@MarDiehl Let's start fresh, revert all of your changes to the meta.yaml and conda_build_config.yaml, and specify the fftw package the following way

requirements:
  host:
    - fftw * *mpi_{{ mpi }}*
  run:
    - fftw * *mpi_{{ mpi }}*

unfortunately, it does not work. Do you have any objection against the last working version? I would then put it into a clean branch or force push (not sure if that works)

[dalcinl]

unfortunately, it does not work. Do you have any objection against the last working version?

Could you clarify? You latest push looks exactly like my last suggestion, plus the # [ mpi != 'nompi'] comment I do not understand why we need it.

[MarDiehl]

unfortunately, it does not work. Do you have any objection against the last working version?

Could you clarify? You latest push looks exactly like my last suggestion, plus the # [ mpi != 'nompi'] comment I do not understand why we need it.

I copied your suggestion and it failed. But currently we have fftw * mpi_{{ mpi }}_* # [mpi != 'nompi'] while your suggestion was fftw * *mpi_{{ mpi }}* which differs not only with respect to the comment but also with respect to double * and trailing _. I'm new to Conda and only have a rough understanding what the code does. Should I try the current version and just strip the comment?

[dalcinl]

I copied your suggestion and it failed. But currently we have fftw * mpi_{{ mpi }}_* # [mpi != 'nompi'] while your suggestion was fftw * *mpi_{{ mpi }}* which differs not only with respect to the comment but also with respect to double * and trailing _.

Oh, sorry, yes! My bad.

I'm new to Conda and only have a rough understanding what the code does.

You are not alone! I also barely know the low level details. It is really frustrating to be a recipe maintainer. I'm truly incompetent. I asked more than once for help (to conda-forge outsiders), but evidently those I asked help for are way smarter than I am.

Should I try the current version and just strip the comment?

Can we please try without the comment? It should not be required, I'm confident it should work. I do not want confusing leftovers in the recipe, because maintenance is already hard for me, not because of time, but because of understanding.

[MarDiehl]

works without the dubious # [mpi != 'nompi'].

[dalcinl]

@MarDiehl Many thanks for your help and patience!

[MarDiehl]

@MarDiehl Many thanks for your help and patience!

Thank you, for the feedstock and for PETSc in general!