search

openjournals / joss

The Journal of Open Source Software
https://joss.theoj.org
MIT License
1.54k stars 186 forks source link

Allow/encourage RRIDs for software refences in text or in the bibliography (as URIs) #427

Open mr-c opened 6 years ago

mr-c commented 6 years ago

See https://github.com/openjournals/joss-reviews/issues/583#issuecomment-406157596 for an example

pjotrp commented 6 years ago

Thanks @mr-c for the idea.

@mr-c just to be clear do RRIDs fixate software versions/releases? I.e., does it work the same way as a Zenodo DOI tagged to releases?

If it is it could be an alternative to using DOIs (eventually) like we do now. JOSS could accept both.

One idea is to accept them now for when there are no DOIs available in the references of a publication - i.e., software published by others the author does not control and where an RRID exists.

mr-c commented 6 years ago

A software RRID is an identifier for the idea of a piece of software, without versioning. It is like saying "XYZ telescope" not "XYZ telescope as it was configured on 2018-07-19". I think they should be listed whenever they exist in additional to the author's requested citation information.

So they aren't sufficient for mentioning a specific version, but they enhance such a listing.

pjotrp commented 6 years ago

Thanks for clarifying. I thinked they missed an opportunity there when (arguably) talking about reproducible research as they do in the first lines on their website:

This is the Resource Identification Portal, supporting NIH's new guidelines for Rigor and Transparency in biomedical publications.

Anyway, not up to us ;)

danielskatz commented 6 years ago

I strongly want to discourage JOSS from accepting RRIDs for this purpose. RRIDs are used in this way in one field (life sciences), but are not used at all in other fields. Instead, I think we need a discipline-independent identifier, such as a DOI, for a software "concept" or package (a set of versions), which is something @mfenner is working on in DataCite and Zenodo is also working on supporting.

mr-c commented 6 years ago

@danielskatz I don't think additional identifiers hurt, so I'm surprised that you think RRIDs should be banned from JOSS. They exist, and are useful, today.

While this came from the life sciences I don't see it as being life-sciences specific as per https://www.force11.org/group/resource-identification-initiative

mr-c commented 6 years ago

Personally I like that RRIDs aren't version specific and that they are easy to distinguish from "point in time" paper/archive DOIs.

danielskatz commented 6 years ago

I don't think additional identifiers hurt, so I'm surprised that you think RRIDs should be banned from JOSS. They exist, and are useful, today.

I believe that using a smaller number of types of identifiers is better than using a larger number of types.

While this came from the life sciences I don't see it as being life-sciences specific as per https://www.force11.org/group/resource-identification-initiative

In what other discipline are RRIDs used for software?

Personally I like that RRIDs aren't version specific and that they are easy to distinguish from "point in time" paper/archive DOIs.

I would rather wait until domain-independent identifiers for software concepts (like the DOIs I mentioned that DataCite and Zenodo are working on) are there than to support a class of identifiers that are only used in one discipline.