testing fix for config variables

[vsoch]

This is a tiny change to ensure that the flux.terminalOutput is correctly passed all the way from the config to the executor! For some reason with task.config, it's not there, likely because it's not known to the task to be parsed?

This was all @pditommaso thank you for the help! Once this is in we are good to go for Nextflow with Flux I think.

• This will close Bug translating nextflow.config nested "flux" variables into workflow #3429

Here is the fixed workflow (before this said the job didn't produce output):

Signed-off-by: vsoch vsoch@users.noreply.github.com

[vsoch]

Okay looks like the session is null (and likely needs to be init somehow)

Condition failed with Exception:

executor.getSubmitCommandLine(task, Paths.get('/some/path/job.sh')) == ['flux', 'mini', 'submit', '--setattr=cwd=/work/path', '--job-name="nf-my_task"', '--output=/work/path/.command.log', '--queue=delta', '/bin/bash', 'job.sh']
|        |                    |     |     |
|        |                    |     |     /some/path/job.sh
|        |                    |     class java.nio.file.Paths
|        |                    id: null; name: my task; type: null; exit: -; error: null; workDir: /work/path
|        java.lang.NullPointerException: Cannot get property 'config' on null object
|        	at nextflow.executor.FluxExecutor.getSubmitCommandLine(FluxExecutor.groovy:66)
|        	at nextflow.executor.FluxExecutorTest.testGetCommandLine(FluxExecutorTest.groovy:71)
<nextflow.executor.FluxExecutor@5655a930 queueInterval=null lastQueueStatus=null session=null name=null monitor=null memoizedMethodClosure$getStageDir=org.codehaus.groovy.runtime.memoize.Memoize$MemoizeFunction@12234922>

	at nextflow.executor.FluxExecutorTest.testGetCommandLine(FluxExecutorTest.groovy:71)
Caused by: java.lang.NullPointerException: Cannot get property 'config' on null object
	at nextflow.executor.FluxExecutor.getSubmitCommandLine(FluxExecutor.groovy:66)
	... 1 more

[pditommaso]

The rescue team is coming ..

[vsoch]

Oh nice! I didn't know there was a rescue team! We have those in my town when people get lost or injured in the mountains. 😆

[vsoch]

I'm seeing a few patterns / ideas that could be promising:

• maybe we need to import nextflow.Session
• we could try a pattern that checks if it's null first (something like session ? session ...) assuming sometimes it's null, other times not.

[pditommaso]

Hold on

[pditommaso]

This is the correct one patch.txt

[vsoch]

Both failed to apply, but I'll look at the changes and see if I can add them to the PR!

[vsoch]

oh it seemed to take! not sure why I got an error message. Should we add something to the FluxExecutor when the session isn't defined, or will it always be defined?

[pditommaso]

No, the session is expected to be defined during normal execution. It's in the test that a fake session needs to be injected

[vsoch]

Ah perfect, thank you rescue crew! 🙌

[vsoch]

Going to bring up my devcontainer environment and be more responsible to actually try the tests before pushing here!

[pditommaso]

Ok, green. Ready to merge it?

[vsoch]

I think so - let me do one more sanity check just to be extra sure!

[vsoch]

Looks great - ship it! 🚢

[pditommaso]

[vsoch]

I come for the code development, and I stay for the animated gifs! 😆