[Obm] DB Abstraction in PHP : Adodb, pear MDB or metabase ?

[Pierre Baudracco]

OBM uses phplib (a customized version) in order to allow access to different 
databases.
As we are going to work on the postgres implementation, we look into database 
abstraction layers.

There seems to be 4 contenders:
- old and venerable phplib (used for low level sessions in OBM too)
- pear MDB
- metabase
- adodb

at first sight I thought phplib was too low level and I had expected other 
modern api would implement a high level SQL abstraction to avoid writing 
different SQL for different DB.
It seems not in fact, except some functionnalities (eg: row limits, date meta 
types...) others still require to write direct SQL.

My first thoughts after some research
- Metabase
pros : meta database description language (in XML) : 1 description for all databases
cons: seems slow

- Adodb
pros: quick, good features set, seems to be the most used, doc
cons: MS ADO inspiration ?

- Pear MDB
pros: what is the ldap driver ?
cons: depend on PEAR (how much ? pear may be good, but i see it as a constraint 
if all pear is needed as it seems to change often), not so quick as adodb


So for now, I'm not sure that migrating the obm abstration layer from phplib to 
other worth the time needed.

does some of you has experimented working with one of these libraries ?
-- 
------------------------------------------------------------------------------
Pierre Baudracco - pierre.baudracco@aliacom.fr
ALIACOM - www.aliacom.fr - Toulouse : 05 62 19 24 91 - Paris : 01 48 25 53 13