Output vs Result

[djarecka]

@tclose - I've been working with pydra this week, and I was having problem on how to get Result object if the task is run in the simplest way as callable. Is it actually possible to get it or I'm able to get only the Output.

This is what I am able to get:

resample = ResampleTask(source_filename=source_file)
output = resample()

resample = ResampleTask(source_filename=source_file)
with Submitter(worker="debug") as sub:
    res=sub(resample)

So the second one gives full Result as expected, but the first only gives output that is equivalent to Result.Output

The problem with this is that we loose the access to many useful informations, e.g., cache_dir, error, etc.

[tclose]

There is a Outputs._results property to access the Results object

[djarecka]

but do we have good reason to hide it?

[tclose]

I was just endeavouring to keep the namespace of the outputs object clear. Although I walked that back a little and added an inputs property to it so you can access the task inputs when building a workflow, so we could also keep "results" or "result" as a reserved output name

[djarecka]

but in general you believe, that task() should not be returning Result?

[tclose]

I could be talked out of it, but I just thought it offers a slightly more convenient/less surprising introduction to Pydra if you get back a "namedtuple-like" outputs object rather than a results object. There is also the option of using the submitter explicitly if you want the result object.

Sorry if this slipped through with all the other changes in my refactor. At this point it would be a bit of a pain to rewrite all the unittests back to handling Result objects, but not insurmountable