[Biopython-dev] Re: BioSQL and PostgreSQL

[Yves Bastide]

Brad Chapman wrote:
 > Yves;
 >
 >
 >>I've adapted BioSQL/*.py to work with PostgreSQL.  Well, at least the
 >>tests pass... Patch attached, comments are welcome.
 >
 >
 > Thanks much! I have committed these. The MySQL still seems to work fine
 > for me, although the PostgreSQL does give me errors in the sql schema I
 > have:
 >
 > IntegrityError: ERROR:  pg_ownercheck: class "bioentry_entry" not found
 >
 >  CREATE INDEX bioentryentry ON bioentry_entry ( entry_id )
 >
 > I know that there is tons happening on the BioSQL development front that
 > I just plain haven't been following, so I am really out of the loop on
 > what the current schema is or what schema we should be working with or
 > if anything is stable.

Hmm, you must have a weird version of the schema :) (this said, the
current one in CVS doesn't work with the corresponding bioperl-db)

 >
 > That being said, if the postgres works for you and MySQL doesn't break,
 > I'm happy :-).
 >
 >
 >>* I've made utility classes for nonstandard SQL/DBAPI operations, e.g.
 >>retrieving the id of the last inserted record.  (The implementation is
 >>so-so, I know :).  There are Generic_dbutils, Mysql_dbutils and
 >>Pg_dbutils.  The choice is made according to the DBAPI module's name;
 >>only MySQLdb and psycopg are currently recognised.
 >
 >
 > Sure. Sounds find to me as this is cleaner then my implementation.
 > Thanks. I am no database expert, so I appreciate anything making this
 > better.

Yep.  But my implementation is a hack: it hardcodes the name of two
adaptors, MySQLdb and psycopg -- while there are at least 3 ones for
postgresql.

 >
 >
 >>(Done but not included in the patch: adaptations to the current
 >>biosql-schema CVS.)
 >
 >
 > Hmmmm, so maybe you understand better than I do what is going on with
 > the current schema :-). I guess the best place to talk about this is the
 > BioSQL list:
 >
 > http://obda.open-bio.org/mailman/listinfo/biosql-l
 >

For discussing about the schema itself: you're right.  But at the
moment, I suppose you would prefer to stick to its old version, since
the new biopython release is slated for next week while I think a new
BioSQL won't come out before December...


 > I wish I knew what was going on so I could offer more advice, but
 > mostly I am a simpleton when it comes to designing SQL schemas so I just
 > let the smart folks figure out what is best and then implement it.
 > Certaintly we do want to stay up to date whenever they have something
 > stable.
 >
 > What do you think about talking up your patches over there?

Hmm, not sure the perl and java guys are interested a lot in biopython's
biosql guts :)
But yes, I've got some questions to ask Hilmar.

 >
 > Thanks again for this work; much appreciated!

Thank you!

 >
 > Brad

yves