[Bioperl-l] identifier interface

[Ewan Birney]

>From my perspective Mat, the LSID spec is just a pretty sensible reworking
of the idea of

  database:identifer.version

 (database <=> namespace)

which has been going around in bioinformatics for a while and seems to
serve people well -- what the LSID spec also provides for is the
java-package like naming system being


  authority/database:identifier.version

where authority is the DNS name and therefore neatly unique. This is a
GoodThing(tm) in my book.


Other things which often "travel with" identifiers in bioinformatics
applications are:

   - display name (swissprot ID vs accession number)

   - description


I would vote in for teh first one, out for the second one


Other things which people can get involved into are things like:


  nested namespaces - more like LDAP - but this nearly always implies some
sort of logical nesting, and we have learnt that biology's logic is nearly
always DAG like and anything "semantic" should be put into an ontology
which links to identifiers...


  security - I personally think this is 100% orthogonal to namespace and
have no idea why they had put it into the LSID spec


  resource management for distributed objects - ditto above




There maybe other things, but the core
authority/namespace/identifier/version is pretty clearly in, so i will
code that part up, as suggested.


Do you have anything else you can imagine putting in?