|
Boost : |
From: Paul A. Bristow (boost_at_[hidden])
Date: 2002-06-26 04:16:59
Many thanks for this prompt response which explains our puzzlement.
I think the moral (with this complex template code) is to avoiid PCH if it
causes a problem,
until we get the next version.
But certainly setting /Zm does not take effect until the compiler is
restarted,
and settings cannot be expected to be the same for all systems.
Paul
> -----Original Message-----
> From: boost-bounces_at_[hidden]
> [mailto:boost-bounces_at_[hidden]]On Behalf Of Jason Shirk
> Sent: Wednesday, June 26, 2002 4:51 AM
> To: boost_at_[hidden]
> Subject: [boost] Re: Re: Problems with uBLAS and /Zm Option (VC7)
>
>
>
>
> "Paul A. Bristow" <boost_at_[hidden]> wrote in message
> news:<AHEJIHEOOOBMJPAGPLIPMEEADBAA.boost_at_[hidden]>...
> > "This should also explain the need for the /Zm option. The compiler
> > needs to ensure it has a contiguous address space for the heap we
> > create the PCH from.
> >
> > Jason Shirk, VC++ Compiler Team"
> >
> > from
> > > http://aspn.activestate.com/ASPN/Mail/Message/1156165
> >
> > I turned off pre-compiled headers for my compilations, just in case.
> > This may explain the different requirements, but not my confusion?
> > Perhaps it is to do with getting the compiler re-started so it finds
> > the contiguous space?
> >
> > (It fails C1060 immediately without appearing to do anything, whereas
> > on the first try, it takes some seconds before failing).
> >
>
> When it fails immediately, the compiler is trying to reserve more memory
> than the system allows. This depends on which version of the compiler
> you have, how much physical memory you've got, and if you've tweaked the
> size of the pagefile.
>
> When if fails after running awhile, an internal heap is exceeding it's
> predetermined limit.
>
> With our upcoming release, this option will rarely be necessary. It
> will never be necessary if you aren't creating a PCH, and it will only
> be necessary if your PCH is > ~50MB (and I'd rethink your build if your
> PCH is that big.)
>
> Jason Shirk
> VC++ Compiler Team
> _______________________________________________
> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk