[fitsbits] DOI keyword usage for FITS? {External}

[Marjolein Verkouter]

Hi all,

>> On 3 Aug 2023, at 21:49, Seaman, Robert Lewis - (rseaman) via fitsbits <fitsbits at listmgr.nrao.edu> wrote:
>> It looks like DOI only works if somebody keeps paying the rent.
>> This is not a limitation I want to find in FITS files.


There's several issues here that I can see:

- The "only works if somebody keeps paying the rent" is only partially correct I'm afraid - I'll explain below.

- someone in the DOI system dropped the ball (IOP by the looks of it). This is not a fault of the FITS system or design nor do I think FITS should try to fix that. The organization dropping the ball should be reminded of their contract they signed with their DOI consortium and make the DOIs resolve again. (see also explanation below).

- any "REFERENC" value in a FITS file that points outside the FITS file itself will be liable to #FAIL and almost by definition will violate the FITS self-describing, self-containing design paradigm

- curious question from me: what is the point of putting a DOI in the FITS file? If the DOI describes the FITS file itself and you are able to read the REFERENC keyword I s'pect you have the actual digital object in hand already. The real question is: what does "the DOI" actually mean? It is totally up to the publisher to define what is meant by "a DOI", and hence, what do you expect to see on the webpage of that DOI once it resolves. Again this is not the responsibility of FITS but of the publisher of the DOI, and later the "owner" or "maintainer" in case the DOIs were transferred to a new owner/maintainer.

But first for the issue at hand: maintaining DOIs after they've been issued should be a fairly low-labour-intensive task, all that's needed is a bit of Python(*) web-based system that:
- its URL can be written in the DOI metadata
- upon opening the DOI's URL the code finds the DOI metdata and renders it on the page
You only need to alter the registered URLs for the DOIs you maintain once - if you need or want to. And keep the webserver running.
This is possibly part of "the rent" that Rob is talking about.

The detailed explanation of "the rent" is the fee model of DOI registration organizations such as DataCite.
Maintaining or issueing DOIs comes with an annual membership fee, and on top of that, for each *new* DOI(**) that is issued in a particular year you get to pay a one-time only cost. Thus, after a DOI is issued there is no "rent" for the DOIs themselves. The costs involved are for the membership, and the obligation to keep a webserver running.

>From our own example: if you manage to join a DOI consortium (which we were able to) the annual membership fee is ~500€ and we run multiple websites- and services anyway.

Which I hope might show that the threshold of "maintaining" might not be as high as one would think, but yes, for sure, it's not zero.

Cheers,
M

(*) or whatever is your favourite web-page-scripting-templating language
(**) I have this on file from the DOI consortium we joined

-----
Marjolein Verkouter (she/her)
Head Technical Operations and R&D

JIVE — Joint Institute for VLBI ERIC —
https://www.jive.eu
Oude Hoogeveensedijk 4, 7991PD, Dwingeloo, The Netherlands
Phone: +31521596516
Mobile: +31625055174
Fax: +31521597332

My working schedule likely differs from yours. Do not feel compelled to read or reply to my email(s) outside office hours.




> I wish we’d thought of this with FPACK.
>   _______________________________________________
> fitsbits mailing list
> fitsbits at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/fitsbits