[ic] Interchange using lots of cpu...

[interchange-users@icdevgroup.org (Kevin Walsh)]

gordon@nextrieve.com wrote:
> >
> > It sounds like something somewhere is copying either the result set or
> > something associated with it with some nasty quadratic behavior.  It
> > wouldn't be Safe.pm would it?  Or some other consequence of firing
> > up perl?  I feel that this is the most likely explanation at this point.
> >
> In fact, when using [benchmark] on nearly every line, it seems that the
> problem is not related to 'firing up perl'.  A single occurrence of
> 	[ARTICLE-param col10]
> for example can take 0.07 seconds during initial loop iterations on 'large'
> result sets.  A bit surprising.  If you want to dump out the content of 10
> fields like that, voila, 0.7 seconds PER ROW, gradually speeding up as
> results are emitted.  The last couple of hundred rows fly by, so unless
> you process more than that you won't notice the problem.
>
I suggest that you set up a simple loop that runs your SQL query and show
[loop-code] for each result.  Benchmark that and then start to add the
rest of your code, a bit at a time.  It won't take long to find the
problem.

> 
> As that seemed so fundamental I gave up on that approach and just dumped
> the [query] results out in a single [perl] tag -- it now runs 100x faster
> at least.
> 
That's what I tend to do for complex loops.

-- 
   _/   _/  _/_/_/_/  _/    _/  _/_/_/  _/    _/
  _/_/_/   _/_/      _/    _/    _/    _/_/  _/   K e v i n   W a l s h
 _/ _/    _/          _/ _/     _/    _/  _/_/    kevin@cursor.biz
_/   _/  _/_/_/_/      _/    _/_/_/  _/    _/