DBD::Pg, DBD::Oracle and others are slow with threads due to DBIS

On Fri, Mar 02, 2012 at 03:11:17PM +0000, Tim Bunce wrote:
> > 
> > The second patch, which makes DBIS more efficient, is a lot more complex,
> > and more likely to break things (especially as it's changing a bunch of
> > macros that are directly #included by the DBD drivers. You may need to
> > bump API version numbers; I don't understand that bit.
> 
> I'm concerned that changing the API, and thus forcing all compiled
> drivers to be recompiled at the same time the DBI is installed,
> isn't worth the gain. Especially as drivers shouldn't be using DBIS in
> any hot code anyway.

I finally got around to looking at DBD::Pg and was horrified to discover
the number of DBIS calls made by hot paths in the code. Most are hidden
behind various tracing macros. Even fetch calls 3 + 3 * num_of_fields!

Then I looked at DBD::Oracle and discovered the same kind of thing.
Which is particularly disappointing for me as I wrote the tracing
mechanism it uses (though maybe that pre-dated thread support).

Anyway, the upshot is that getting DBIS calls out of major drivers will
require a fair bit of work. It's just grunt-work really, nothing complicated.
As you noted with DBD::mysql, Dave, the performance gains are worth it.

Any volunteers? Naturally I'll be happy to suggest an approach I think
will work well (basically an extension of that used by DBD::Oracle).

Dave, I'll get back to you about the DBIS changes soon, I hope.

Tim.