Yes, I agree, but it is in fact not possible to search it in under 3 seconds.
Once cached in RAM, I can do a fgrep -r in under 800ms.
That's still slow, but our webserver isn't exactly the fastest thing in existence.
> Bumping the timeout on an inefficient bit of code doesn't fix the actual
> problem, it just slows down other queries.
Ok, then does that search code not work at all.
It works, but just doesn't scale.
Perhaps http://www.pmwiki.org/wiki/Cookbook/FastSearch, or we forget
it and users search theirs keywords under google :)
I think not - putting 1.2MB on grep's command line can't end well. Shelling out also bypasses max_execution_time.