On Mon, Jul 28, 2008 at 12:44 PM, Yang Tse <yangsita_at_gmail.com> wrote:
> 1) The underquoted definition of AM_PATH_GTK in line #7 of
> /opt/csw/share/aclocal/gtk.m4, It could be this but currently it
> doesn't seem very likely, other autobuilds survive other underquoted
> definitions.

Yep.

> 2) A problematic /opt/csw/bin/gm4 GNU m4 version 1.4.5 which, as logs
> show, is actually failing with a 'memory exhausted' error. No other
> autobuild is using 1.4.5 But other autobuilds, IRIX ones, are even
> using plain 1.4 version and don't fail with this.

I think we can rule this one out, I tried with version 1.4.11 freshly
built from original source and it fails the same way.

> 3) A system wide or per process memory configuration limit is killing us.

It doesn't seem likely though, I ran the test with:
-bash-3.00$ ulimit -a
core file size (blocks, -c) unlimited
data seg size (kbytes, -d) unlimited
file size (blocks, -f) unlimited
open files (-n) 1024
pipe size (512 bytes, -p) 10
stack size (kbytes, -s) 8192
cpu time (seconds, -t) unlimited
max user processes (-u) 8005
virtual memory (kbytes, -v) unlimited

(I raised -n from 256 to 1024 with no difference)

> 4) A somehow corrupted autom4te.cache Not very likely but we could try
> to let buildconf remove it, just to protect ourselves.

I removed the cache (and all build-created files/dirs), but it still failed.

> 5) Strange inherited ACLOCAL_FLAGS. Buildconf could be modified to show this.

I'm no autoconf expert, but if you tell me how I can probably dig out the info.
[,,,]
> Until this issue is sorted out, it would really help to have the
> Solaris autobuild running more frequently than once a day. ;-)

I can fix that, no problem.

-Tor
Received on 2008-07-28