Karsten Hilbert wrote: > Can we agree on this: > > 1) migrate to psycopg2 > > Why ? > > - actively maintained/developed
Just to point out that Gerhard Haering finally released pyPgSQL 2.5 recently on SourceForge, with all the outstanding patches which we required in it, followed by v2.5.1 to accommodate the recent PG security fixes relating to Unicode and SQL injection attacks. None of that is to suggest that GNUmed may not be better served changing to psycopg2, just to point out that pyPgSQL is once again being maintained in a satisfactory fashion. Tim C > - lets us avoid mx.DateTime > - offers connection pooling (reduce gmPG complexity) > - has more unit tests now > - offers async queries > - offers access to SQLSTATE error codes > > 2) use SQLSTATE based error handling > > - do not rely on textual (possibly i18ned) error messages > > 3) implement Ian's concept of exception-throwing business objects > > - drop SELECT FOR UPDATE but > - use XMIN in update operations in > - serialized transactions > > Karsten _______________________________________________ Gnumed-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnumed-devel
