Subject: Re: compiler-related problems building lang/python2.2
To: None <tech-pkg@NetBSD.ORG>
From: kent <>
List: tech-pkg
Date: 05/08/2002 22:08:55
On 8 May 2002 16:10:00 -0500, Frederick Bruckman wrote:
> On Wed, 8 May 2002, Greg A. Woods wrote:
>> => Removing rpath references to buildlink directories.
>> ===> Building for python22-2.2.1
>> cc -c -DNDEBUG /opt -I. -I./Include -DHAVE_CONFIG_H  -o Modules/python.o Modules/python.c
>                  ^^^^
> There's an argument missing. With two arguments lacking flags, it
> looks to gcc like you're specifying two sources. Probably, the command
> contains ${SOME_VARIABLE}/opt, where ${SOME_VARIABLE} is supposed to
> start with "-I", perhaps, but ${SOME_VARIABLE} is not defined.

configure creates the cc options list, and it appears that something there
is misfiring.  My Makefile ends up with:

OPT=            -DNDEBUG -g -O3 -Wall -Wstrict-prototypes

I just built it again:

===> Building for python22-2.2.1
cc -c -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -I. -I./Include -DHAVE_CONFIG_H
 -o Modules/python.o Modules/python.c
cc -c -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -I. -I./Include -DHAVE_CONFIG_H
 -o Parser/acceler.o Parser/acceler.c

Two questions.

1) Is there a std way to tell configure *not* to use -g here? I
don't see an option to turn it off.

2) I am experiencing some stability problems with NetBSD Python
2.2.1 that didn't exist with earlier versions. In the least damaging
case, I'm getting a Bus error, and in the worst case, it completely
locks the machine up. It always locks the machine up if I'm running
Noatun when it happens. :^(   I don't have a good way to reproduce
this problem because I can only get it to happen with a particular
set of scripts which recursively walk a series of directories and
do extended data consistency analyses on several dozen moderately
large datasets.  It also happens at different times in the analysis
sequence, so doesn't appear to be data related. (I've also never
before seen a bus error with any Python code in the 8 years I've
used Python)

Again, this is only with Python 2.2.1 and only on NetBSD that this
happens. The same scripts run fine on Solaris.  Anyone else seen
anything like this?