[MOBY-l] G2G as a moby service

[Jason E. Stewart]

Hey All,

I've put up a preliminary G2G page at:

  http://genex.sf.net/G2G.shtml

It really doesn't have much useful at this point, other than a link to
the slides I used at the O'Reilly conference.

Here are some issues:

* G2G will need a directory server, like moby proposes, as well as a
  way to describe what services any given server provides. So we see
  no need to reinvent the wheel, and instead I'd rather use moby's
  support for WSDL and UDDI, and/or help write it.

* I wonder about SOAP. Because we already have MAGE-ML as a markup
  language for gene expression data, SOAP doesn't really buy us
  anything. I'm not sure that it increases our overhead, either, but
  I'm curious what y'all think.

* complex services/queries: we'll be using the MAGE object model to
  specify queries for gene expression data. So it's not likely that
  we'll have seperate services for things like 'what species do your
  experiments investigate' and 'what list of PI's created your public
  experiments'. Instead we imagine a query mechanism based on the OQL
  - object querly language. So we'd I imagine we'd break up our
  services into how much of the OQL syntax a given server implements,
  or something along those lines.

I mentioned it in my talk at O'Reilly, but in case others missed it,
I'd like to hear whether folks think the MAGEstk infrastructure will
be useful for other data types besides the MAGE work. For those
unfamiliar with it, MAGEstk enables us to take a UML object model (in
the case of MAGE for microarray gene expression data) and
automatically derive the following pieces:

* XML transport layer
* API's in Java/Perl/C++
* Reader/Writer softare to convert between XML <==> API objects

Eventually we'll finish the final piece which enables:

* rDBMS persistence layer
* software to convert between rDBMS <==> API objects

More info can be found at mged.sf.net

Cheers,
jas.