This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Less good ;-) results from Rittle's string allocator
- From: Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>
- To: libstdc++ at gcc dot gnu dot org
- Cc: pcarlini at unitus dot it
- Date: Wed, 28 Nov 2001 02:49:59 -0600 (CST)
- Subject: Re: Less good ;-) results from Rittle's string allocator
- Organization: Networks and Infrastructure Lab (IL02/2240), Motorola Labs
- References: <3C03962D.A893E65D@unitus.it>
In article <3C03978F.431C4DC0@unitus.it> you write:
> Of course, in the second set of results, N=400 and M=20 ;-)
No problem. I don't know why I admit this in public but I crashed my
machine *twice* yesterday looking at your test case before I
correlated the problem and knew it was easily repeatable. ;-)
Somehow, one just doesn't notice that a short test case will
eventually attempt to allocate 2^400 bytes (of course, it dies long
before that point on most hardware due to sizeof size_t, the fact that
total swap space requirements may well exceed that ever to be
available on Earth, etc ;-).
This is on a machine that *never* crashes. Why did it crash? I
learned a new rule today: "Don't allow unlimited-sized core dumps on a
memory file system."
Regards,
Loren