On Sun, 4 Sep 2011, Tom Lane wrote:
> What I would suggest is to see whether a more recent x86 version shows
> the problem or not. If not, let's just write it off as an already-fixed
> compiler bug.
I have installed the most recent version in the home directory of a
purpose-made user on that machine.
configure:3252: icc --version >&5
icc (ICC) 12.0.5 20110719
Copyright (C) 1985-2011 Intel Corporation. All rights reserved.
git checkout 6416a82a62db4e66b2edb0fa8fc83a580c3f1931
to get a revision I knew was right in the broken range for mongoose.
Apparently they deprecated one of my compiler flags: -xN (N is for
Nocona), seems they renamed it to -xSSE2. Since this is a one-off run, I
ignored that warning.
The result is no crash in the cube test.
I think tomorrow I'll try to get the 9.0 compiler set up on a clean VM,
and if the issue duplicates there, I can see about setting up SSH access
if anyone is still interested in investigating this further.
In response to
pgsql-hackers by date
|Next:||From: Marti Raudsepp||Date: 2011-09-05 07:52:44|
|Subject: Redundant bitmap index scans on smallint column|
|Previous:||From: Oleg Bartunov||Date: 2011-09-05 06:39:30|
|Subject: Re: WIP: SP-GiST, Space-Partitioned GiST|
pgsql-committers by date
|Next:||From: Bruce Momjian||Date: 2011-09-05 13:45:09|
|Subject: Re: Re: [COMMITTERS] pgsql: Remove "fmgr.h" include
in cube contrib --- caused crash on a Ge|
|Previous:||From: Tom Lane||Date: 2011-09-05 00:07:51|
|Subject: pgsql: Guard against using plperl's Makefile without specifying--with-|