From: Phillip Seaver (phil_at_[hidden])
Date: 2005-09-20 11:52:02
Kevin Wheatley wrote:
> However something good has come of it, I know that naming lots of
> files the same thing is probably not a good idea for Boost build - not
> something I'd be likly todo except in automatically generated
> tests/etc. Suggests some kind of hash map or tree search followed by
> a linear list data structure maybe?
It's since been fixed, which is one of the reasons I mentioned the
earlier email. When I modified his python scripts so that all the file
and class names were unique, the build times were within seconds of each
other (unique vs. lots of the same). As you said, not likely to happen
in real life, I think.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk