[Biocorba-l] Using the Naming Service as a top level object

[Brad Chapman]

Hey all;
Thanks for the feedback.

Juha:
> > The use of naming service is very good idea!

Alan:
> If what we need looks like a Naming Service, and acts like a Naming
> Service. Then using (or at least looking!) at the Naming Service would be
> a good idea. 

I'm glad to hear I wasn't way out of my mind on this one. Yeah, I
was starting to miss BioEnv (at least a top level factory-like
object), but it does seem like most of the functionality is just
"What databases do you have?" and "Give me a database," which we may
be able to do as easily with Naming Services.

> Who knows, maybe we'll even start to think about directory
> and trading services (c.f. UDDI, SOAP, LDAP, OmniGene).

Hey now, don't get out of control here :-)

Juha:
> > To make it more useful we need also consistent naming schema as you
> > wrote. Your naming schema looks good. Please have a look what we have:
> > 
> > http://corba.ebi.ac.uk/cgi-bin/ns/ns.start
> > 
> > E.g: databases/EMBL/nsdb/Embl

Alan:
> Personally, I've never been happy with parts of this particular naming
> scheme, e.g. the inclusion of "EMBL" seems both redundant and
> inappropriate.

Your naming scheme (and the one used in AppLab) seem on target with
my thinking. My only comment is that the names seem too deep. I know
java coders seem to like deep nested org.blah.spam.foo.bar.whatevers
before their libraries, but this makes it harder to ask a question
like: "What datatabases are available?", since you have to traverse
all these heirarchies.

So we have:

databases.BioCorba/db_name.db_version
vs.
databases/db_name/db_part/name_of_server

Could we settle for something in the middle like:

databases.BioCorba/db_category/db_name.db_version
analysis.AppLab/analysis_category/analayis_name.analysis_version

This is kind of what AppLab does with its categories, and
seems like enough depth for reasonable resolution and
differentiation. What do you guys think?

Alan:
> If we're going to do this, can I make a plea to do it in reasonable
> bite-size chunks and "KISS", please? I've had enough of trying to work on
> over-arching all-inclusive schemes...

I'm in 100% agreement with you here. All I'm interested is really is
a "standard" way to add BioCorba objects to the Naming Service so we
know how to retrieve them. It's good for it to be extendable to
other non-BioCorba cases, but I'm not interested at all in figuring
out how these other cases should name their databases, etc. I'll
leave it up to Ewan to come up the high-level names and interfaces
for all potential biological objects :-).

Brad
-- 
PGP public key available from http://pgp.mit.edu/