[wp-hackers] Google SoC: PostgreSQL port

Brian Layman Brian at TheCodeCave.com
Fri Mar 23 18:04:01 GMT 2007

> take a look on http://pear.php.net/package/MDB2

I'm mostly talking to myself here as I haven't really followed this thread
or stared into the black, black eyes of wp-db.php, but YOU brought up data
abstraction layers again.  And IF we were to consider putting in a complete
data abstraction layer beyond what we already have (and I don't think that's
really what you were saying), I think ADOdb offers a more proven track
record with large open source projects (such as ACID, PostNuke, Xaraya,
phpWiki, Mambo, PHP GACL, TikiWiki, eGroupWare and phpLens App Server) than
what MDB2 has.  

ADOdb:  http://en.wikipedia.org/wiki/ADOdb  http://adodb.sourceforge.net/

However, I've not used MDB2 with anything yet either.  So, perhaps I'm not a
fair judge.  

Still the benchmarks do show that ADOdb has significantly better performance
than MDB:

Benchmark was to select 82 rows from the products table 200 times.
The tests were repeated 5 times. Connection times were excluded
from the benchmark. Lower values are better. All numbers are in 

MySQL	 1.12 	1.12	1.17	1.15	1.14
ADOdbext 1.30	1.31	1.29	1.30	1.32 
dbx ext	 1.35	1.38	1.41	1.37	1.36	(index only)
ADOdb	 1.43	1.47	1.47	1.44	1.45
dbx ext	 1.53	1.52	1.52	1.52	1.55	(index/assoc/info)
PhpLib	 1.53	1.62	1.64	1.64	1.57   
MDB      1.77	1.75	1.75	1.76	1.73
PEAR DB	 2.91	2.90	2.85	2.83	2.84	(fetchInto)
PEAR DB	 3.14 	3.13	3.22	3.12	3.16	(fetchRow)
M'base	 4.51	4.55 	4.46 	4.54 	4.52    (numeric indexes)
M'base	 4.99	4.72	4.71	4.71	4.72	(assoc indexes)

Plus, while ADOdb does provide support for PostgreSQL, I'm biased to it
because I have a client that wants their WordPress system to be moved over
to Interbase.  So, I'd love to see some progress made on that front, as just
I might end up canoeing that creek all by my lonesome self, with or without
a paddle.  (Of course a non-core side project wouldn't have to deal with all
the broken plugins...)

If we were to do a SOC thing and have someone paid $4.5 grand for three
months of work (or whatever it is), putting in a layer that allowed WP to
run on 18 different database systems would be a much more appropriate and
universal project.  

Of course, any project of this size would require buy in from a number of
mentors and should involve a high level of communication with the plugin

Brian Layman

More information about the wp-hackers mailing list