[Pipet Devel] gui protocol revisited

[J.W. Bizzaro]

Brad Chapman wrote:
> 
> I thought that we could keep both the streaming XML dialog *and*
> CORBA, but just in two different places in the program. My idea was
> that we could keep the small middle that we have in Loci right now to
> communiate with the GUI front via the streaming XML dialog you
> describe.

Okay, if you like that.  And I guess Jarl likes it too.

>     This small middle could then also implement the CORBA interfaces
> to pass the information created in the GUI (as a work flow diagram)
> into the "processing" portion of the program. It would make sense to
> pass this as DOM or XML, since it seems like a nice structured way to
> store the data (and since Overflow already uses it to feed its "pull"
> networking system). I think that eventually the processing part could
> move through a work flow diagram in an even "smarter" way (ie. based
> on the fastest way to implement it) which is what Jarl seems to be
> proposing for GMS's with the neural net and genetic algorithms
> processor. Here it seens to makes sense to start with what we have
> (the Overflow pull system) and move into something more robust as
> things move along.

I'm a little confused about this 'processing part of the program'.  Is this
something GMS or Overflow has now?  Is it something you'll have to write? 
Also Brad, recall how Loci was going to use an XML database (XDBM) to store
the workflow diagram (network and subnetworks).  How does this relate to the
'processing' part?  Perhaps you can explain this to the others and how we plan
to use XML.  Will this be replaced by a DOM structure?

>     I think this would make it easier to implement addition front ends
> (since they would only need to deal with XML though a pipe, and not
> implementing a CORBA interface), and will also allow us to re-use the
> code we already have for dealing with communicating with the GUI. In
> addition, I think this will allow us to make a "simpler" CORBA
> interface between the small middle and Overflow and GMS program
> processing functions, since we won't have to worry specifically about
> GUI requirements while coding these.

Okay.

>     Does this make any sense? I think this gives us a lot of reuse of
> code, while maintaing the streaming XML that Jeff digs and the
> CORBA/API interfaces of Overflow and GMS.

I was just thinking that if we're using CORBA at all anywhere, why not use it
everywhere?

Cheers.
Jeff
-- 
                      +----------------------------------+
                      |           J.W. Bizzaro           |
                      |                                  |
                      | http://bioinformatics.org/~jeff/ |
                      |                                  |
                      |        BIOINFORMATICS.ORG        |
                      |           The Open Lab           |
                      |                                  |
                      |    http://bioinformatics.org/    |
                      +----------------------------------+