Svanberg Liss wrote: > > This confuses me a bit. Are you saying that you don't need a DB, or that you > don't want to store the contents of the DB in a DB-specific manner? We need a DB. I'm just saying that the elements stored in the DB are really separate files on a filesystem. So, the DB would behave something like a file manager or even a file system. > ( I believe that you wan't to be able to send queries to the DB such as " > give me all interfaces/objects/applications that can preform this operation > upon this kind of data "? > Isn't that a DB? ) Yes. We want to be able to do that. But I think our database should allow for a query mechanism to be plugged in as a separate locus. I alluded to this a little earlier. Here is how a query would be set up: DATABASE LOCUS -----> QUERY LOCUS -----> DATA LOCUS > Hmm. > There is a Gnome project called the Gnome-DB ( you didn't expect that name, > did you? :) > http://www.chez.com/rmoya/gnome-db/index.html > > and it has got a DB-engine that uses raw files. I have not examined it yet ( > save that I have read their homepage ) so I don't know much about it, but it > might be worth a look. That should be worth looking into. > If you pass around CORBA objects, that "represents" the actual file, you > will not need to duplicate the file. > ( GNOME::Stream is a very good candidate! :) EXACTLY! > In either case, passing around a file "reference", instead of reading > directly from the file, seems to be the obvious solution. YES! :-) Cheers. Jeff -- +----------------------------+ | J.W. Bizzaro | | jeff at bioinformatics.org | | | | THE OPEN LAB | | Open Source Bioinformatics | | | | http://bioinformatics.org/ | +----------------------------+