[fitsbits] REFERENC keyword, etc.

[THIERRY FORVEILLE]

> Does your e-mail render this as a link?
> 
>  	doi:10.7717/peerj-cs.1
> 
> It's only 'future-proof' in that it's broken now, so when it's broken in
> the future, no one will notice that anything has changed.  It'd be nice if
> worked in e-mail readers, PDF makers, and the like but the majority of
> tools have no idea that they're even supposed to do anything with it.
> 
> That's why the library community recommends a URL:
> 
>  	https://dx.doi.org/10.7717/peerj-cs.1
> 
> I admit that there are some issues with disabiguation, as all of the
> following resolve to the same place, so it's not ideal as an identifier:
> 
>  	http://dx.doi.org/10.7717/peerj-cs.1
>  	https://dx.doi.org/10.7717/peerj-cs.1
>  	http://doi.org/10.7717/peerj-cs.1
>  	https://doi.org/10.7717/peerj-cs.1
>  	http://doi.org/6q5
>  	https://doi.org/6q5
>  	http://dx.doi.org/6q5
>  	https://dx.doi.org/6q5
> 
> ... but if the goal is to get people from the short string of text in the
> FITS file to the actual document, an HTTP or HTTPS URL is better than a
> DOI URI.  Even if their viewer doesn't automatically make the URL
> clickable, our current generation of scientists have enough familiarity
> with seeing 'http://' all over the place that they know what to do with
> it.
> 
> If we come up with a standard that only calls for URIs in the form
> 'doi:...', I'll recommend that our community do:
> 
>  	REF_INST='doi:10.xxx/whatever'
>  	COMMENT   http://dx.doi.org/10.xxx/whatever
> 
> ... or something similar.
> 
> If we're talking about archiving, we should be worrying about making sure
> that the information is usable to both current & future generations, not
> worrying about shaving a few bytes off a file.  If we were concerned about
> the size of the headers, then we'd come up with standards to store the
> information so it's normalized to at least 3NF.
> 
I must say that I have a hard time figuring why a line in FITS headers would
need to be clickable in a browser or mailer. I mean, it's not as if either
browsers or mailers were the primary tools used for FITS display/analysis ;-)