On 11/26/2010 12:07 AM, Bruce Dubbs wrote:
> DJ Lucas wrote:
> 
>> The update to expect broke jhalfs. I
>> haven't looked into why yet, but last time I had seen similar error, it
>> was due to a missing role= tag on one of the screen blocks. 
> 
> I suspect it's because the package was renamed expect5 instead of 
> expect-5.  IIRC, the same problem existed with tcl8.
> 
> I think the problem may be in common/urls.xsl where an option needs to 
> be added similar to the tcl option at line 87.
> 
> For a work around, try editing jhalfs/lfs-commands/chapter05/040-expect
> and s/$PKGDIR/expect5.45/
> 

Work around I used was directly in the Makefile.  I added the missing
lines to the 040-expect target directly and it finished without issue
AFAICT. I had the exact same thing happen once before when adding
initd-tools to my local copy and I didn't use either the role= or remap=
tags in the instructions. I just reviewed expect.xml and it looks
correct, so you are probably correct about the real problem.

Still dealing with sort issue. No hardware issues found. I need to
isolate the issue either to this machine or to a certain environment,
unfortunately, I don't have another 64bit multi-core machine at my
disposal. So far, I've managed to limit it to UTF-8 on multi-core
(possibly only 32bit on a 64bit multi-core CPU). Doesn't make sense for
this to show up on a system, that has thus far been stable, after such
significant changes to the code (added multi-threading support to sort)
and it be a hardware issue, especially when sidestepping the new code
doesn't display the problem. Reviewing the strace output now...

Any chance you (or anyone) could test the following in a 32bit userland
on 64bit multi-core system? If a segmentation fault occurs, then it's
not my system...and if it does, then I need to do a more thorough
hardware test.

zcat cracklib-words-20080527.gz | LANG=en_US.UTF-8 sort -u

Removing the i18n patch resulted in no change.

-- DJ Lucas

-- 
This message has been scanned for viruses and
dangerous content, and is believed to be clean.

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to