[Pipet Devel] Re: [Pipet Devel] Re: which handles distribution (was: vsh?)

[J.W. Bizzaro]

Brad Chapman wrote:
>
> I think it's important that Overflow and GMS aren't completely
> dependent on each other (so you could use GMS for "big" programs
> with distributed systems and use Overflow for "small" libraries on a
> local scale).

I agree that there should be some separation of functionality, but to the
point where it would still be possible (with code modification) to get one
application to execute without the others being present.  IOW, I still don't
want to see GMS or Overflow dismantled, dissolved and absorbed.

BUT, if our projects are to merge, none of them will any longer be stand-alone
applications.  Of course.

> However, I know for my needs that I would like to use
> both "small" and "big" together, and so I'd like to see Overflow and
> GMS interoperate, so that you don't have to choose one core over
> another.

Right.

> Would it be difficult to define a "distributed API" for GMS
> that Overflow could plug into (when distribution over multiple
> computer was necessary)? Then Overflow could define a "small library
> API" that GMS could plug into to run libraries? Does this make any
> sense?

Yeah, I think this has to be done.  Neither Overflow nor GMS can be agnostic
about the other.  But, for Loci's sake, we don't want to be flip-flopping on
how user-to-core-to-application/object communication will occur.

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