Benno Schulenberg wrote: > There's still one coreutils po file for which 'msgfmt -c' reports a > fatal error, but I can't see anything wrong with the indicated > messages. It might be a bug in 'msgfmt'. > > coreutils/be.po:2298: number of format specifications in 'msgid' and > 'msgstr[0]' does not match > coreutils/be.po:2306: number of format specifications in 'msgid' and > 'msgstr[0]' does not match > msgfmt: found 2 fatal errors
msgfmt is right: When you look at the plural formula of this file, you see that each of the three forms can occur for infinitely many numbers n. coreutils/be.po:2298 has this: #: src/dd.c:542 #, c-format msgid "1 truncated record\n" msgid_plural "%<PRIuMAX> truncated records\n" msgstr[0] "1 абрэзаны запіс\n" msgstr[1] "%<PRIuMAX> абрэзаных запісы\n" msgstr[2] "%<PRIuMAX> абрэзаных запісаў\n" Here the msgstr[0] presents wrong information for n = 21, 31, ... because it pretends that n = 1. Yes it is subtle, and I don't expect that translators would notice this by themselves; this is why msgfmt does the check. Bruno
