Re: [Gc] test.c patch
ivmai at mail.ru
Thu Dec 10 22:43:24 PST 2009
"Boehm, Hans" <hans.boehm at hp.com> wrote:
> I committed the patch.
> I agree that NO_INCREMENTAL is redundant. The GC_DISABLE_INCREMENTAL environment variable is a better solution. I'm all in favor of removing it after 7.2 is out.
> NO_INCREMENTAL existed in 7.1. Thus I don't think there's a benefit to getting rid
> of it before 7.2. It was always a debugging option. And it made a bit of sense
> before GC_DISABLE_INCREMENTAL, and possibly before the pseudo-environment-file
> support you recently added, but no more.
Ok. Let's postpone (though I don't see really the reason of keeping it since it existed in 7.1). Not a big deal.
> > -----Original Message-----
> > From: gc-bounces at napali.hpl.hp.com
> > [mailto:gc-bounces at napali.hpl.hp.com] On Behalf Of Ivan Maidanski
> > Sent: Wednesday, December 09, 2009 11:22 PM
> > To: gc at napali.hpl.hp.com
> > Subject: Re: [Gc] test.c patch
> > Hi!
> > "Boehm, Hans" <hans.boehm at hp.com> wrote:
> > > Unless someone can think of a reason not to, I'll shortly
> > check in the attached patch:
> > >
> > > * test.c (main, WinMain): Consistently don't invoke
> > GC_enable_incremental
> > > if MAKE_BACK_GRAPH is defined, but do invoke it with
> > parallel marking.
> > >
> > > This should remove some of the unexpected platform
> > differences in heap size..
> > >
> > > Hans
> > > ATTACHMENT: application/octet-stream (test.c.diff)
> > 1. I don't mind.
> > 2. I think NO_INCREMENTAL should be removed completely
> > because the macro means something like "build the collector
> > with the incremental support but skip testing it (because
> > it's probably broken)".
More information about the Gc