add example roc metadata

Add a draft for example roc metadata for the fixed script example

requested review from @weil

assigned to @caroott

Wouldn't ./workflows suffice here?

We can change it to that

@wetzels, the @id should be relative to the root of the RO-Crate, right? Is it with an ./ or without?

I prefer the variant including ./, as it makes it clearer that it is relative path. I don't think it's necessary though.

Regarding the first question: ./ should be the root directory of the ARC. What does the arc subdirectory do there? Looks weird to me, so I agree.

Yes, i also agree. This was a relic from the arc mount in the cwl file

changed this line in version 3 of the diff

Also F#?

F# is already referenced in the workflow. We could reference it in both, or just the part where it is specified that it is F#

Is this runs or a specific run, aka something like runs/ColumnAddition?

Yes, in this basic example it was just "runs" so far. But naming it makes more sense

Hey, so in this example, the defined workflow and run have nothing to do with each other, or am I missing something?

added 1 commit

• c611caf2 - fix missing #

Compare with previous version

Yes, i also talked with Timo about that we need to connect them

Alright!

But I think we could also keep them as separate examples, right? One run->workflow combo and one stand-alone run?

Of course in this case it would also make sense to actually have this structure in the ARC itself. So the RO-Crate json and ARC would be synchronous.

@muehlhaus

Do you mean separating them in two files or having one run without any workflow? As it is, a run can't stand by itself, since it needs the workflow it executes as well, unless we duplicate all the information

I meant having two runs in the ARC and then (naturally) also in the RO-Crate. One for the case where it executes a workflow and one where it's just a script execution, without any workflow defined.

Just judging from the specification, this should be a valid case. But I see that we wouldn't need an example for it if it's kind of discouraged approach.

I think we could reference the workflow used in that run here (hasPart), what do you think?

Not sure whether hasPart would be okay here, given it being used to depict filesystem structure in RO-Crate. (Although I agree that it would make sense)

An alternative would be about. This way the relationship assay -> process would be equivalent to run -> workflow.

@wetzels would you also add it to the about section or do you think hasPart could be ok?

I need to have a closer look into this. But first: I would disagree that assay->process is equivalent to run->workflow. Can you elaborate on this @weil?

Now some general question: what are you describing here? To me, it looks like an ro-crate for the workflow/run part of an arc only. Everything else is missing, right? Did you use any of the existing work on workflow-run profiles for ro-crate (https://www.researchobject.org/workflow-run-crate/)?

Whats currently described here is the general workflow in the arc, and the specific inputs used to create the result in the runs folder of the arc. The profile describing the workflow can be found here (CWL), and the execution part is based on that profile.

added 1 commit

• 2dcf2c1f - make ./ and not ./arc root folder

Compare with previous version