[BUG] Component header and Freetext Value cause exception

[omaus]

Describe the bug
ValidateAgainstHeader causes an error in everyday usecase.

To Reproduce
Steps to reproduce the behavior:

1. Create new ARC
2. Fill out investigation metadata
3. Create study and fill out metadata
4. Test pipelines, should pass
5. Add new metadata sheet to study and add Component building block
6. Test pipeline, should succeed
7. Add a row to your Component building block
8. Test pipeline, pipeline fails with metadata error
9. Change component to something else, e.g. Characteristics
10. Pipeline succeeds

Expected behavior
Everyting's fine.

OS and framework information (please complete the following information):

• OS: Win10
• OS Version: Pro 64-bit (10.0, Build 19045)

Additional context
Full history of the bug occurence.

[HLWeil]

So Pipeline does not flatout fail

Just added a study with a component column with two values.

[HLWeil]

Or is this the Issue:

[SabrinaZander]

The overall pipeline seems to pass, but when you have a look at the details it looks like that.

The Publication Metadata fails and then you cannot publish the ARC.

[HLWeil]

Ahh so the Invenio package needs to be referenced?

[SabrinaZander]

Yes, sorry, that's missing in the steps to reproduce.

Maybe just for the sake of completeness. We had an ARC that we wanted to publish. The Invenio validation package had to be added for this. Then this metadata error occurred in the pipeline. The ARC could not be published because the ARChigator page always showed the error "Could not load metadata".
According to the error description in the pipeline, it has something to do with the Swate building blocks. After a lot of testing, I found out that it was the component building block. I can't say exactly where the error came from. But after I replaced all the component building blocks, the error disappeared and the ARC could be published. See here again:

nfdi4plants/arc-validate#112

I can also invite you to some test ARCs if you want to.

[HLWeil]

Okay I see. I could reproduce the error now after referencing the Invenio Test package and looking into it 👍

[HLWeil]

But another requirement for this to fail is, that the value in the cell needs to be freetext. Terms work just fine.

[HLWeil]

Okay this is already fixed in ARCtrl, should suffice to update the dependency in arc-to-invenio

[HLWeil]

Okay so the Issue should fixed.

@SabrinaZander @Hannah-Doerpholz would be great if you could confirm

[SabrinaZander]

Sorry, but now the error looks like this. Or do I need to do something else besides restarting the pipeline?

[HLWeil]

Maybe you could invite me to the ARC @SabrinaZander, so I can have a look?

[Hannah-Doerpholz]

Since I have the same issue in my test ARC I invited you to that one if that helps. There is no "secret" data in that one @HLWeil

[HLWeil]

Okay after some more investigation, it seems like this is not a problem of any of the source code.

Instead, the docker packaging of the tool does not work as intended, i.e. calling the tool fails.

Which also becomes apparent in the image you posted:

[HLWeil]

Should be fixed with 0f4fba8

@Hannah-Doerpholz, I restarted your pipeline and it ran through. @SabrinaZander, could you also check?

[Hannah-Doerpholz]

It ran through for my actual "data" ARC as well. Thank you!

[SabrinaZander]

Thank you very much, in my test ARC the pipeline runs successfully.
In another ARC, in which I first encountered the metadata pipeline problem, there is still an error.
Now called internal error:

If you would like to investigate this further, I would be happy to invite you. It is no longer relevant for me, as we have already remade the ARC and published it.

[HLWeil]

Yeah, I'll take a look!

[HLWeil]

Okay, this problem is actually a mismatch between the strictness of the validation and the conversion, regarding the information in the contacts:

1. Validation passes if at least one person has all required fields*
2. Converter requires all persons to have the required fields

In this case, one person was missing the affiliation property.

So, in principle the goal will be to align both tools to follow the same rules. I don't want to decide on which of the two approaches to take by myself, so I'll move this to a discussion.

---

*Required fields

• First Name
• Last Name
• E-Mail
• Affiliation

[HLWeil]

https://github.com/orgs/nfdi4plants/discussions/10