Houghton,Andrew writes:
 > > From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of
 > > Eric Hellman
 > > Sent: Wednesday, April 01, 2009 9:51 AM
 > > To: [log in to unmask]
 > > Subject: Re: [CODE4LIB] registering info: uris?
 > > 
 > > There are actually a number of http URLs that work like
 > > http://dx.doi.org/10.1111/j.1475-4983.2007.00728.x
 > > One of them is http://doi.wiley.com/10.1111/j.1475-4983.2007.00728.x
 > > Another is run by crossref;  Some OpenURL ink servers also have doi
 > > proxy capability.
 > > So for code to extract the doi reliably from http urls, the code needs
 > > to know all the possibilities for the doi proxy stem. The proxies also
 > > tend to have optional parameters that can control the resolution. In
 > > principle, the info:doi/ stem addresses this.
 > 
 > Again we have moved the discussion to a specific resolution mechanism,
 > e.g., OpenURL.  OpenURL could have been defined differently, such
 > that rft_id and rft_idScheme were available and you used the actual
 > DOI value and specified the scheme of the identifier.  Then the issue
 > of extraction of the identifier value from the URI goes away, because
 > there is no URI needed.

Yes, that would have been OK, too.  But no doubt there are other
contexts where it's possible to pass in an identifier without also
being able to say "and by the way, it's of type XYZ".  Surely you
don't disagree that it's good for identifiers to be self-describing?

It's the same with actionable URLs: isn't it better than I can tell
you:

	http://www.miketaylor.org.uk/dino/pubs/

Instead of having to say:

	www.miketaylor.org.uk/dino/pubs/
	Oh, by the way, access this using HTTP rather than FTP.

 _/|_	 ___________________________________________________________________
/o ) \/  Mike Taylor    <[log in to unmask]>    http://www.miketaylor.org.uk
)_v__/\  "A Linux system requires rebooting about as often as a Windoze
	 system requires re-installing" -- David Joffe.