|
Boost-Build : |
From: Franz (fschnyder_at_[hidden])
Date: 2007-09-27 08:24:18
Vladimir Prus writes:
>
> Franz wrote:
>
> > Hello
> >
> > Has anybody ever investigated on the -j bug 'The -j flag can cause bjam to
> > get confused when single actions update more than one target at a
> > time...'?
> >
> > My Problem is, that our development machines are updated to dual cores and
> > a -j2 would cut down the building time to nearly half the time used with
> > -j1. But we can't use -j2, because we have actions, were we hit the -j bug
> > and the build always fails the first time, so we switched back to -j1.
> >
> > I had a quick look at the jam source code but I'm quite lost in it. I
> > found the execcmd and execwait functions but I couldn't yet figure out how
> > jam waits for the end of a action. Has anybody an idea were I could start
> > looking for the bug, or could give me hint how the actions are executed
> > and waited for? Thanks for any help. Id would be nice to use the full
> > power of our new dual core machines.
>
> Do you have a pointer to description of that bug?
http://www.boost.org/doc/html/jam/miscellaneous.html#jam.miscellaneous.bugs__limitations
-> 'The -j flag can cause bjam to get confused when single actions update more
than one target at a time. bjam may proceed as if the targets were built even
though they are still under construction.'
- Franz
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