Closed cboettig closed 5 years ago
This is fixed. What do you think of pulling the version from that file if it's available? So in an APA citation for this package it would include (Version 0.0.9) after the package name.
Nice. Yes, I think pulling the version as makes sense (could be done from the DESCRIPTION file too).
On the codemeta parsing, it might also be good to check first for a referencePublication
field (though maybe that should be renamed something more intuitive like suggestedCitation
or citeAs
see https://github.com/codemeta/codemeta/issues/144). The name
field etc will always refer directly to the SoftwareSourceCode object, whereas this field is a ScholaryArticle
type can describe a paper associated with the publication instead (identical to what CiteAs very nicely does when it finds CITATION files)!
On including versions, as you no doubt know, including versions is considered best practice, e.g. by the Force11 working group paper, but I'm not particularly bullish on it myself. I agree that version information is important, but the problem is that there's no way to also capture precise versions of the dependencies, since the software package itself won't state these -- by design, most software is compatible with multiple versions of it's own dependencies.
Whoops, my apologies, I was a bit out of date on terms. Looks like the standard practice is to use citation
, not referencePublication
for this purpose, and you will already find the citation
in most codemeta.json files of R packages using CITATION, e.g.:
https://github.com/ropensci/RNeXML/blob/master/codemeta.json#L387-L449
Ok the software now pulls the version from codemeta.json. Also, as you mentioned if a citation field is present, it will use the fields in that part of the document as the input to be parsed.
http://citeas.org/cite/https://github.com/ropensci/piggyback finds the codemeta.json file, but parses it incorrectly; using the DESCRIPTION as the title. (should use "name", since
schema:name
isowl:sameAs
dc:title
).