From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-02-12 10:49:29
David Abrahams wrote:
> >> If you mean the code which processes the #include... lines, using the
> >> header cache, it would be called much less.
> > Do you mean that the same file is scanned several times in common
> > setting? Why?
> Normally, yes, because stock Jam is stupid ;-).
I don't see much improvement for now.
> I'm not sure if the header cache patch by itself eliminates this
> re-scanning, but it looks like it does. However, turning on the
> HCACHEFILE will speed up consecutive builds by preventing re-scanning
> between builds.
Well, the problem is that the rule which handles includes will be called
anyway, and that rule takes a lot of time. I'll have to optimize that rule.
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