more on cat-gets/tst-catgets

[Jack Howarth]

     I now have confirmation that this failure of catgets/tst-catgets
is cross-platform effecting other arches in debian as well. Tonight
I compared a build of Jakub's glibc 2.2.92-2 srpm with my current
2.2.93-1 debian package. What I noticed was although Jakub's srpm
passed the catgets/tst-catgets test on a redhat based ppclinux machine
there was a distinct difference in the de.msg generated in the
catgets subdir. In particular a diff between the redhat copy and the
debian copy is of the form...

--- de.msg.rh	2002-09-03 23:46:06.000000000 -0400
+++ de.msg.debian	2002-09-03 10:34:15.000000000 -0400
@@ -22,1010 +22,979 @@
 $ msgstr "DATENDATEI [AUSGABEDATEI]"
 $ 
 $ Original Message: 
-1 \
-Project-Id-Version: GNU libc 2.2.5\n\
-POT-Creation-Date: 2001-08-17 13:03-0700\n\
-PO-Revision-Date: 2002-06-06 09:30+0200\n\
-Last-Translator: Karl Eichwalder <ke@suse.de>\n\
-Language-Team: German <de@li.org>\n\
-MIME-Version: 1.0\n\
-Content-Type: text/plain; charset=ISO-8859-1\n\
-Content-Transfer-Encoding: 8bit\n
+
 $ : locale/programs/ld-monetary.c:278
 $ , c-format
 $ Original Message: \t\t\t\t\t\t\t      %s: value for field `%s' must be in range %d...%d
-2 \t\t\t\t\t\t\t      %s: Wert für Feld »%s« muss im Intervall %d...%d sein
+\t\t\t\t\t\t\t      %s: Wert für Feld »%s« muss im Intervall %d...%d sein
 $ : nis/nis_print.c:277
 $ Original Message: \t\tAccess Rights : 
-3 \t\tZugriffsrechte : 
+\t\tZugriffsrechte : 
 $ : nis/nis_print.c:275
 $ Original Message: \t\tAttributes    : 
-4 \t\tAttribute      : 
+\t\tAttribute      : 
 $ : sunrpc/rpc_main.c:1427
 $ , c-format
 $ Original Message: \t%s [-abkCLNTM][-Dname[=value]] [-i size] [-I [-K seconds]] [-Y path] infile\n
-5 \t%s [-abkCLNTM][-DName[=Wert]] [-i Größe] [-I [-K Sekunden]] [-Y Pfad] Eingabe
datei\n
+\t%s [-abkCLNTM][-DName[=Wert]] [-i Größe] [-I [-K Sekunden]] [-Y Pfad] Eingabeda

The redhat gettext is 0.11.4-3 and the debian is 0.11.5-1. They appear to
be roughly identical with any patches applied in the redhat gettext
already in the 0.11.5 tarball. Also we see this issue with gettext
0.10.40 as well. Any ideas were to go from here to backtrace this
and find out why the numbering is left off of the de.msg generated on
debian?
                                 Jack
ps I do notice that file catgets/de/libc.cat is dramatically different
in size between the debian and redhat builds...

-rw-r--r--    1 root     root        13138 Sep  3 10:34 libc.cat.debian
-rw-r--r--    1 root     root        71938 Sep  4 00:06 libc.cat.redhat