[DAS] Re: [Call to action] Retrieval of positions from feature identifiers

[Thomas Down]

On Wed, Nov 28, 2001 at 11:32:53AM +0000, Ewan Birney wrote:
> On Tue, 27 Nov 2001, Brian Gilman wrote:
> 
> > I totally agree with Mathew on this one. Each "Business Entity" can be
> > easy mapped to into this domain. Why reinvent the wheel?? Our group will
> > be deploying a UDDI server for people to play with. I think this will
> > allow us to get our head wrapped around UDDI and how it fits in the web
> > services architecture. 
> 
> 40 SOAP messages to do discovery?

Nah, not that I can see...

In the UDDI version 2.0 Programmer's API Specification
(http://www.uddi.org/), I count 10 API calls for
inquiry/discovery operations.  There are then a further
16 which are only used by code which wants to insert data
into a registry.

It's a relatively fine-grained API -- no overloaded operations.
Doesn't look excessive (or hard to implement/use) to me.

In any case, I'd sincerely hope that we'd be using it safely
wrapped away behind some simple interfaces.  For Java applications,
JNDI (Java Naming and Directory Interfaces) might be the sensible
solution.

> (I have to admit, all this SOAP/UDDI whatever stuff *is just badly written
> CORBA* Jeez. People want a Trading service right from CORBA terms?)

Horses for courses.  I still like SOAP for the case where you
want to move large volumes of structured data (feature tables,
or whatever) over the network.  CORBA seems a bit limited in what
it can do here, especially once you start avoiding valuetypes, anys,
etc.

In any case, we should be concentrating more on data-model
and query semantics far more than transport.  Since I started
using the Axis toolkit, I've been quite impressed by the extent
to which I've been able to ignore SOAP.  Toolkits are definitely
maturing quite quickly now.

    Thomas.