Zenodo seems like a Good Thing. Should Zenodo be configured to automatically mint a DOI based on Github releases? It seems like this is a simple way to handle things since no manual intervention is needed.
Seems like the last release on Zenodo is 2.11.0, https://zenodo.org/record/58873
It’s enabled. The latest version is 2.14. I don’t know where you found that
link – how do we update that info?
See: https://zenodo.org/account/settings/github/repository/stan-dev/stan
The badge from the github readme points to the right version:
Could I give you the reigns? Maybe you want to create ones for RStan,
CmdStan, PyStan, and Math separately?
I searched for “Stan” on zenodo and only found the 2.12 one. Strange!
I did too and I also found that one. I don’t know how to update Zendoo to point to the latest.
I turned on pystan but I don’t have permission for rstan and cmdstan and
math. I think you can just flip the switch(es) on the following pages
(just as it currently is for stan-dev/stan) and all will be well:
@ariddell, I enabled it for CmdStan, RStan, and Math. Is that the right thing to do? (I don’t think there’s any harm in it.)
@bgoodri, please let me know if you want me to disable RStan.
@bgoodri or @jonah, please let me know if you want me to enable anything else. (or do it yourself?)
Zenodo says I don’t have permission
Try now?
Still says permission required.
I hate permissioning… I’ll try a few more things. brb.
I’ve got nothing. Can you log out and log back into Zenodo? Or refreshing?
I made you an admin on stan-dev’s GitHub.
I think it is working now.
Stan github page https://github.com/stan-dev/stan/ has now link to Zenodo page https://zenodo.org/record/221370#.WG-cCPHfquU which says
Cite as:
Daniel Lee, Bob Carpenter, Peter Li, Michael Betancourt, maverickg, Marcus Brubaker, … Daniel Mitchell. (2016). stan-dev/stan v2.14.0 [Data set]. Zenodo. http://doi.org/10.5281/zenodo.221370
which strangely has “[Data set]” in the recommended citation.
Aki
Do you know how to fix that?
Sounds like a bug. In other news, I can’t search for Stan past v2.12.0:
https://zenodo.org/search?page=1&size=20&q=stan&type=software&sort=mostrecent
Just make sure to add this to the dev ops time sheet!
Every time we add things like this, it leads to downstream
maintenance cost.
I don’t have an opinion about whether DOIs are worth the
effort.
+1. I’m gonna put up a post on the dev-only section regarding what keys I
have.
The dev-ops roll, I can keep updating in a public post.
The signal advantage of this system is that it just creates DOIs
whenever you make a github release. There should be zero time spent on
this once it’s working.
ariddell Developer
January 6
The signal advantage of this system is that it just creates DOIs
whenever you make a github release. There should be zero time spent on
this once it’s working.
That’d be nice, but I wouldn’t bet the farm on
Zenodo maintaining a consistent API or even existing
long term (what happens if Thomson-Reuters buys them
and rebrands, for instance?). And who knows what
GitHub’s going to do in the future.
If we add this to all our interfaces and packages, it’ll
require some work whenever a new package is added to get
the scripting working.
Plus, we should have some minimal doc on how it works on
stan-dev to make future dev ops easier.
The point I’m trying to make is that we should go into
these things with the realization that there will be
some long term effort involved. Nothing’s free that’s
meant to persist. I was just saying I don’t have an opinion
on whether DOIs are worth the effort. The Stan project’s
intentionally set up to give you the flexibility to do it
if you think it’s worth the effort. Otherwise, we’d be
mired in endless bureaucracy or turn into a dictatorship.
The fun part about the project for me is being able to improve
it in ways I think are worthwhile while getting improvements
from everyone else coming in at the same time.
Bob, I think we’re broadly in agreement. In general I don’t bank on any
new service being around for more than a year or three.
Zenodo seems like it might well stick around. It seems like its backing
is more or less comparable to the backing enjoyed by arXiv.