Citation

[murphyqm]

Should the citation file point to the published paper or the zenodo "cite all versions" doi?

Leaving it as is for now (pointing to paper), but can update to point at zenodo doi in minor update if needed... @andreww does this sound sensible?

[andreww]

Difficult to know to be honest. Ultimately the person writing a paper who has used the package is probably best placed to decide what to cite and how to cite it and all we can do is provide useful pointers.

From the point of view of a hypothetical future user who has made use of the variable physical properties stuff and run some models using pytesimal with a fancy new core, what they should probably do is say something like:

"we treated the mantle with temperature dependent thermal conductivity (Murphy Quinlan et al. 2020)"

and

"Calculations were performed using version 2.x.x of the Pytesimal package (Murphy Quinlan et al. 2021) using our modified treatment of the planetesimal core"

That 2020 citation would then be to the JGR paper. The "2021" reference could be to a future JOSS paper or to the Zenodo DOI. In some sense JOSS exists to allow citations to software in cases where a reference to a repository DOI is not included (either because the author doesn't want to cite that kind of thing or because the journal does not allow it) and citing a paper that is essentially a wrapper around the software somehow easer.

So this leads on to what we should do to make the life of this hypothetical future user easer, and encourage best practice. I think the approach taken in https://github.com/ConorMacBride/mcalf#citation (a recent JOSS paper was on this) is pretty good but it does not reference back to JOSS (and I'm not sure if it is technically possible to do so). Copying essentially the same information (but also with a list of e.g. BibTeX items) into a CITATION.txt file probably makes sense. This leaves the question of what to do with a .cff. I think this ought to be a citation to the code itself, so to the Zenodo DOI, but a message saying something like "further information can be found in the README/CITATION file may be of use.

Maybe leave this issue open while we think further about it?