interesting performance article

[Martin Pool]

   http://www.acmqueue.com/modules.php?name=Content&pa=printer_friendly&pid=366&page=1

Nothing world-shatteringly new, but worth a read anyhow... particularly
the apparent contradiction between "don't prematurely optimize" and
"don't leave performance til the end", and the importance of benchmarks
that are repeatable, observable, easily presented, realistic, etc.

-- 
Martin