[Cdk-devel] Molecular database structure

[Ola S. <ola...@fa...>]

Hi,

Sorry for cross-posting, but this concerns both the CDK and Bioclipse
projects.

I am looking at how the molecular database-plugin for bioclipse (bc_db)
should be designed, and I have some thoughts and proposals that I would
like comments on.

1) We use Hibernate3 so that we get transparent OORDB-mapping and
multiple database support (and follow the standard of EJB3)

2) We provide the functionality to organize molecules in collections
(folders)

3) We store XML in a relational database as text (I know, but it IS much
faster)

4) We provide HSQLDB with biolipse as an embedded database

5) Connected database-instances show up in the FileViewer (that possibly
should be renamed?)

6) A user can do everything with a database object that he can do with a
file object now

7) The database supports subsearches, fingerprint-searches,
inchi-searches etc.

Comments? How much is already done in NMRDB? What about XMLDB? We could
provide another plugin to support that, I suppose. What would we gain
from using an XMLDB instead of an RDB?

Well, just some thoughts. A database-plugin is probably a very nice
feature of Bioclipse.

Cheers,

   .../Ola