Performance measurement and optimization

>   The question is: Is it slow by design, which'd be a pity since I like it,
>or is just the implementation slow? It'd be really great if someone could
>have a look at improving the speed of PEAR/DB. This has not be Stig, who is
>currently very busy. This would be an issue where some of the people from
>the PHPLIB mailing lists could step in: go through the sources of PEAR/DB
>and make suggestions. And, no, by suggestions I don't mean replacing PEAR/DB
>with PHPLIB/DB_Sql :-)

Does anyone have a feel for where the time is going? In my experience the 
best way to look at these things is to ...

Find a fairly simple set of code that exhibits a problem you want to solve 
(bad performance, bug, etc).

Use profiling, tracing, or other performance testing tools to isolate the 
problems or design flaws or language bugs, etc.

Come up with a candidate solution, and implement it.

Iterate until the system is fast enough for your needs.

I'm speaking as a fairly new user of php; it might be interesting to do 
some profiling of some standalone php code tho to see where the time is 
going. If someone had some code exhibiting the bad behavior (I've not seen 
it yet myself; my php apps are still pretty simple) I might have some time 
to have a stab at this.

--bob