Open tillea opened 5 years ago
Hi Andreas,
Thanks for your comment. I think I remember that doi2cff
only covers zenodo dois ( or maybe even Zenodo software dois? Not sure). Also when looking at the specific error that indeed seems to be the cause.
When I do
doi2cff init https://doi.org/10.5281/zenodo.2609141
I get something that looks like valid CFF, based on the metadata from the corresponding Zenodo record.
Hope this helps, -Jurriaan
Hi Jurriaan, thanks for your quick reply.
On Tue, Jun 25, 2019 at 08:41:14AM +0000, Jurriaan H. Spaaks wrote:
I think I remember that
doi2cff
only covers zenodo dois ( or maybe even Zenodo software dois? Not sure). Also when looking at the specific error that indeed seems to be the cause.
I admit I had no idea that Zenodo dois exist and specifically what might be the difference to any other DOI.
I get something that looks like valid CFF, based on the metadata from the corresponding Zenodo record.
Does this mean other dois then Zenodo will not supported? It would help to stress this more explicitly and provide a better error message than a bug trace output from the script.
Thanks for this software anyway, Andreas.
So the difference between dois from various sources is that they potentially have metadata in a different format, which doi2cff
then must translate to CFF. Since we (the Netherlands eScience Center) use Zenodo to assign dois to releases of software that we create, @sverhoeven thought it useful to have a tool that helps people with creation of a CFF file based on whatever metadata can already be harvested from (in our case) Zenodo; with the option to extend the software for other sources/other doi issuers, like for example FigShare.
Note that the doi2cff
workflow assumes that the metadata from Zenodo is good. And that is also its main problem, because the metadata is usually not good unless authors really make an effort, either by manually editing the metadata for their records, or as I myself tend to do, by including a .zenodo.json
file in my repos (basically a way to tell Zenodo what you want the metadata to be).
Because of this, I personally prefer to have the reverse flow direction of information, so I use the following workflow:
cffinit
cffconvert
to convert the CFF information to a .zenodo.json
file (other formats are available as well)git add CITATION.cff .zenodo.json
, commit and push to the repomake a new release
on github, I'll get a persistent identifier (doi) from Zenodo.With this workflow, I get good metadata on Zenodo, I get a DOI, and I get CFF which is consistent with whatever the metadata is on Zenodo.
As for your question, Does this mean other dois then Zenodo will not supported?: if I may speak for @sverhoeven: we are not against creating/incorporating converters that convert from e.g FigShare dois to CFF, it's just that it is not near the top of our todo list. We'd welcome Pull Requests that extend or improve doi2cff
's functionality, but note that this is generally only useful when you have a whole bunch of records for which the metadata is good, and for which you want to create CFF files.
Hi again, thanks for the clarification. I assumed a general DOI parser which is probably a naive assumption. So please add a better error handling for those DOIs you can not parse. Kind regards, Andreas.
It may be better to use the DataCite metadata as that would provide support for other data repositories (e.g. Figshare). But regarding journal article DOIs, this should be possible using the Crossref API.
Description
I'm trying to build a Debian package from doi2cff. The usage example
works. However, if I try random other DOIs I get:
Same happens for
Any idea what might went wrong for these DOIs?
Kind regards, Andreas.