[feat] Print more information about dependencies in detailed test listing

[ekouts]

Fixes #1521.

Some output examples:

- OSUAllreduceTest_4:
    Description:
      OSU Allreduce test

    Environment modules:
      <none>

    Location:
      /users/karakasv/Devel/reframe/tutorials/deps/osu_benchmarks.py

    Maintainers:
      <none>

    Node allocation:
      standard (4 task(s))

    Pipeline hooks:
      - post_setup: set_executable

    Tags:
      <none>

    Valid environments:
      gnu, pgi, intel

    Valid systems:
      daint:gpu

    Dependencies (conceptual):
      OSUBuildTest

    Dependencies (actual):
      - ('OSUAllreduceTest_4', 'daint:gpu', 'gnu') -> ('OSUBuildTest', 'daint:login', 'gnu')
      - ('OSUAllreduceTest_4', 'daint:gpu', 'intel') -> ('OSUBuildTest', 'daint:login', 'intel')
      - ('OSUAllreduceTest_4', 'daint:gpu', 'pgi') -> ('OSUBuildTest', 'daint:login', 'pgi')

- OSUDownloadTest:
    Description:
      OSU benchmarks download sources

    Environment modules:
      <none>

    Location:
      /Users/karakasv/Repositories/reframe/tutorials/deps/osu_benchmarks.py

    Maintainers:
      <none>

    Node allocation:
      standard (1 task(s))

    Pipeline hooks:

    Tags:
      <none>

    Valid environments:
      gnu

    Valid systems:
      daint:login

    Dependencies (conceptual):
      <none>

    Dependencies (actual):
      <none>

I kept the old Dependencies because the check itself might have dependencies to more checks than will appear on this specific system.

[pep8speaks]

Hello @ekouts, Thank you for updating!

Cheers! There are no PEP8 issues in this Pull Request!Do see the ReFrame Coding Style Guide

Comment last updated at 2020-11-07 15:01:35 UTC

[ekouts]

@jenkins-cscs retry all

[codecov-io]

Codecov Report

Merging #1576 (4c7b707) into master (8982bdb) will decrease coverage by 0.00%.
The diff coverage is 89.47%.

@@            Coverage Diff             @@
##           master    #1576      +/-   ##
==========================================
- Coverage   88.02%   88.02%   -0.01%     
==========================================
  Files          45       45              
  Lines        7016     7030      +14     
==========================================
+ Hits         6176     6188      +12     
- Misses        840      842       +2     

Impacted Files	Coverage Δ
reframe/frontend/cli.py	79.75% <89.47%> (+0.21%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8982bdb...4c7b707. Read the comment docs.

[vkarak]

What I can't understand from the output samples you posted is why the "Dependencies for current system" are empty for the second set of examples, when you have for example test dependencies.

[teojgo]

What I can't understand from the output samples you posted is why the "Dependencies for current system" are empty for the second set of examples, when you have for example test dependencies.

Executing this:
./bin/reframe -C unittests/resources/settings.py --system=sys0 -c unittests/resources/checks_unlisted/deps_simple.py -L

looks good for me.

[ekouts]

What I can't understand from the output samples you posted is why the "Dependencies for current system" are empty for the second set of examples, when you have for example test dependencies.

@vkarak The environments e0 and e1 are not in the configuration, so no test is going to run in the end

[vkarak]

@vkarak The environments e0 and e1 are not in the configuration, so no test is going to run in the end

This is a bit problematic and comes from the fact that the printing is based on tests not on the test cases. I was about to comment something related to that for your PR. Perhaps we can address also this problem. If the tests are not going to run, they should not be listed.

[ekouts]

This is a bit problematic and comes from the fact that the printing is based on tests not on the test cases. I was about to comment something related to that for your PR. Perhaps we can address also this problem. If the tests are not going to run, they should not be listed.

Okay, I can easily print only the checks that are included in the index_of_deps. Or do you prefer a different way to solve this issue?

[vkarak]

I don't like the index_of_deps concept being computed outside and then just being passed along all the way down to format_checks(). The interface doesn't make sense. The list_checks() should take the test cases instead and derive the actual tests and dependencies from there. You build the dependency index privately inside that function or wherever it is actually needed.

[ekouts]

Okay, I think I can pass only the testcases and infer the checks from there. This way we will only have the checks that will run

[vkarak]

I did some small changes. Lgtm now.