Posted by NC on 02/17/06 05:42

J wrote:
> "NC" <nc@iname.com> wrote in message
> news:1140057031.204786.34080@g44g2000cwa.googlegroups.com...
>
NC> Last times folks at phpLens tested it, PEAR:db had an
NC> overhead of 150-170%:
>
NC> http://phplens.com/lens/adodb/
>
> Those benchmarks are interesting. Is it PEAR itself, or just
> a biproduct of OO design i.e. lots of classes to create, more
> function calls, etc?

I don't know for sure, but the effects are clearly there, aren't they?
I could theorize that the part of the overhead that disappears when
accelerators are used is largely a by-product of bloated code base
(things like class definitions that are not used to instantiate objects
within a particular script), while the part of the overhead that
remains is the penalty imposed by the code that is actually used, but I
have no idea if this theory is any good...

Cheers,
NC

[Back to original message]