Recent update of coreutils yielded an unexpected result on my box. Namely, the Polish gettext catalogue (pl.po) is broken beyond repair. OTOH coreutils-5.95 have the translation correct.
The description of the problem follows: * In 5.95, pl.po file contains a header entry: "Content-Type: text/plain; charset=ISO-8859-2\n" and the file indeed is in the ISO-8859-2 encoding. So, each of 18 locale-specific characters is encoded in one byte. Example (in hex): ł (l with /) => B3 ę (e with ,) => EA ą (a with ,) => B1 ó (o with ') => F3 * In 5.96 pl.po file contains the same header, but *ALL* Polish-specific characters were magically transformed into a three-byte combination: ł (l with /) => EF BF BD ę (e with ,) => EF BF BD ą (a with ,) => EF BF BD ó (o with ') => EF BF BD Additionaly: # file coreutils-5.95/po/pl.po pl.po: ISO-8859 PO (gettext message catalogue) text # file coreutils-5.96/po/pl.po pl.po: UTF-8 Unicode PO (gettext message catalogue) text This is a real problem, because the file encoding is not recoverable without lots of manual tweaking... Help? -- Registered Linux user #390131 (http://counter.li.org)
_______________________________________________ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils