Red Hat Bugzilla – Bug 40656
Bad multibyte chars in japanese .po-file
Last modified: 2014-03-16 22:20:47 EDT
[xmms-1.2.4-15.src.rpm] Description of Problem: xmms fails to rebuild because msgfmt fails on po/ja.po. How Reproducible: everytime Steps to Reproduce: 1. LANG=C rpm --rebuild xmms-1.2.4-15.src.rpm Actual Results: ... file=./`echo ja | sed 's,.*/,,'`.gmo \ && rm -f $file && PATH=../src:$PATH /usr/bin/msgfmt -o $file ja.po ja.po:1675: invalid multibyte sequence ja.po:3110: invalid multibyte sequence ja.po:3167: invalid multibyte sequence ja.po:3212: invalid multibyte sequence ja.po:3265: invalid multibyte sequence found 5 fatal errors make[2]: *** [ja.gmo] Error 1 make[1]: *** [all-recursive] Error 1 make: *** [all-recursive-am] Error 2 ... Additional Information: $ rpm -q gettext glibc gettext-0.10.37-2 glibc-2.2.2-10 $LINGUAS is unset
Hm, it builds fine with the 7.1 gettext. Trond, has gettext changed significantly in this regard?
The strings it fails on are input strings. These are in ISO-8859-1 and does indeed contain sequences that are not legal EUC-JP.
Yes, gettext has changed significantly wrt. handling of Japanese (multibyte) - to use the old style, use OLD_PO_FILE_INPUT=yes
This should be fixed in the current rawhide xmms (1.2.4-16)