[Pipet Devel] An apology, Loci, gnome, and CORBA

Justin Bradford justin at ukans.edu
Wed Jun 23 19:30:57 EDT 1999


> When I start reading the mailing list, I'm appaled to find out that we need to 
> have network aware apps, and workflow, and Gateways and Hubs, and PAOS, and 
> lion and tigers and bears, oh my!

It's really not as bad as it looks. Really.

> The things we've discussed make for a huge system. Paos and the locid in 
> particular look nasty. This is where I almost turned back and decided to write 
> my own simple shell using unix pipes and simple stuff like that.

I'm currently at Genentech in the Biocomputing department, and that's
basically where things are now (except with a web front-end and a sprinkle
of Java). I'd like Loci to be a next generation kind of project, but
really, I think we have more complicated interfaces, but not a whole lot
more programming than doing it the old way.

> We could reinvent the wheel in developing loci, but it's best if we at least 
> know what everyone else's wheels look like first. We've got most of the hard 
> parts of loci done, let's try building loci over the top of the gnome CORBA 
> layer, and if it's too slow then consider (a fixing ORBit, or b rewriting just 
> the parts of CORBA we need).

As we are making use of GNOME, I would agree that it makes sense to use
their technology for these things. It makes all of the tools more
flexible in the GNOME environment.

However, there has to be some flexibility in the interface (to the data in
particular). What if I want to make a tool that uses data in a unique way
(which we don't have an interface for)?

I guess, fundamentally, I'm interested in Loci from a data storage,
management, and interface issue. I'm thinking of a system which can pull
data from various sources in a very flexible way and then combine and
present that data in an equally flexible way to any arbitrary program.

Justin





More information about the Pipet-Devel mailing list