I said that a more robust system would require only the
specification of the DOI that would be used to retrieve automatically
all the metadata associated with the work (I understand that this is
possible, although I never tried it).
Yes, it's true. Every DOI is a URL like http://dx.doi.org/10.123456/foo.bar.
If you fetch it up normally it redirects you to the document, typically a
page at the publisher with the abstract and a link to buy it, or in our case
links to the RFC in various forms.
If you fetch it and ask for "Accept: text/bibliography; style=bibtex" or
application/json, you get the bibliographic data in bibtex or JSON. It's
pretty handy.
R's,
John