Last rts changes break
Thu, 9 Aug 2001 09:55:48 +0100
> 9 Aug 2001 06:34:04 GMT, Marcin 'Qrczak' Kowalczyk=20
> <email@example.com> pisze:
> > Well, it's the second (not third) full build cycle after cvs update,
> > so I'm not 100% sure that the bug is present in the current version.
> > Perhaps it's in the previous version (the first build produced a
> > defective compiler and possibly good libraries, in the second build
> > the bad compiler crashed). I'll try to bootstrap from 5.00.2.
> Strange. I thought that bootstrapping three times from 5.00.2 is
> a long time, so I stoped, made maintainer-clean, and started again
> using the version that previously crashed. I thought that if the
> bug has been fixed during last changes, I'll try to compile this
> file with different options, perhaps -Onot, and it will be fine:
> the compiler produced would be OK and I would have it quickly.
> But this time it didn't crash. So the bug is non-deterministic and
> rare. I still don't know if it's present in the *last* version.
There was a GC bug in the compacting collector yesterday (there may
still be more, of course). The compacting collector was also enabled by
default yesterday, now it isn't.
Are you running with a different GHCRTS than before, perhaps? (sigh, I
guess I'm going to be asking this question a lot from now on.)