Bug 52783 - kdebase does not build due to meinproc error
Summary: kdebase does not build due to meinproc error
Alias: None
Product: Red Hat Raw Hide
Classification: Retired
Component: kdebase   
(Show other bugs)
Version: 1.0
Hardware: All
OS: Linux
Target Milestone: ---
Assignee: Bernhard Rosenkraenzer
QA Contact: Ben Levenson
Depends On:
TreeView+ depends on / blocked
Reported: 2001-08-29 01:26 UTC by Christian Hechelmann
Modified: 2007-04-18 16:36 UTC (History)
0 users

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2001-09-03 01:15:00 UTC
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
Quick hack to force LC_NUMERIC in meinproc(1) to POSIX (615 bytes, patch)
2001-09-03 21:49 UTC, Christian Hechelmann
no flags Details | Diff

Description Christian Hechelmann 2001-08-29 01:26:48 UTC
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)

Description of problem:
build of kdebase 2.2-5 bombs in doc/kate when make runs meinproc in 

Version-Release number of selected component (if applicable):

How reproducible:

Steps to Reproduce:
1.rpm --rebuild kdebase-2.2-5.src.rpm
2. bombs in doc/kate

Actual Results:  
[drdisk@junkyard kate]$ meinproc --check --cache index.cache.bz2 
Don't know how to do adjust column widths with libxslt
unable to parse index.docbook

Expected Results:  successful build

Additional info:

The error messages comes 
from /usr/share/apps/ksgmltools2/docbook/xsl/html/table.xsl:157 which is 
part of kdelibs-2.2-5

relevant packages installed at time of build:


Comment 1 Bernhard Rosenkraenzer 2001-08-29 09:56:31 UTC
I can't reproduce this here and it works in the build system as well.
The only difference I see is that you're using an ancient version of glibc.

Comment 2 Christian Hechelmann 2001-08-29 20:12:07 UTC
glibc version does not seem to matter, it breaks at the same point on 2
different machines (one based on roswell, the other RH7.0 + much of current 

One interesting thing: meinproc does only barf on docbook files containing
<tgroup cols="some-value">

Namely kate/index.docbook, konsole/index.docbook, kwrite/index.docbook.

The error itself comes from the <xsl:choose> block 
in /usr/share/apps/ksgmltools2/docbook/xsl/html/table.xsl around the lines 141-
166, which explicitly checks for the xslparser and takes specific action 
depending on the parser found. I did a quick search for adjustColumnWidths, but 
didn't find any match in libxslt, libxml*, docbook-*.

Will do some further investigation.

Comment 3 Christian Hechelmann 2001-09-03 01:14:55 UTC
Now we're getting somewhere:

I did a fresh reinstall of roswell on a spare box (workstation type + KDE + 
"meinproc --check /usr/share/doc/HTML/en/kate/index.docbook" still breaks.

The locale was set to de_DE@euro by the installer.

I then did some testing with various locales. I unset LANG, and all of LC_
and then called meinproc with "LANG=xxx meinproc --check ...".
The results were:

Fails for: cs_CZ,de_DE,de_AT,en_DK,no_NO,pt_BR
Works for: cs,da,de,en_GB,fr,gl,it,ja,ko,nl,no

Further investigation yielded LC_NUMERIC as the culprit, it seems that
any locale that has a decimal_point other than '.' makes either meinproc or 
libxslt break. This seems to be the same old bug that plagued netscape: It 
produced broken postscript output for any locale with something different than 
"." as the decimal point.

Comment 4 Bernhard Rosenkraenzer 2001-09-03 13:54:17 UTC
Argh, locales...
Fixed in 2.2-10

Comment 5 Christian Hechelmann 2001-09-03 21:25:26 UTC
There's no point in fixing kdebase, since the root of the problem, usage of %f 
or %g format specifiers in *printf()/*scanf() without caring for l10n, is 
either in kdelibs (meinproc) oder libxml2/libxslt. My bet goes for libxml2 
and/or libxslt. Fixing it in kdebase only cures the symptom, not the cause.

Comment 6 Christian Hechelmann 2001-09-03 21:49:38 UTC
Created attachment 30734 [details]
Quick hack to force LC_NUMERIC in meinproc(1) to POSIX

Note You need to log in before you can comment on or make changes to this bug.