23 Apr 2006 (Sun)

11:16:36 # Life manpage and their encoding. I've had a brief review of manpage and their encoding. Although it is possible to have a manpage coded in UTF-8, it is currently somewhat impractical to do so. man-db searches for files based on the locale string, in the order of ja_JP.UTF-8, ja_JP, ja (if you take Japanese as an example), which means that the locale string really needs sanitization (case-sensitiveness is required here), and files coded in different codesets than the current one will not be searched for. What 'ja' means is defined in src/encodings.c of man-db. Some good documentation is in order. UPDATE: According to GyrosGeier, the encoding-specific directories are there only for cases where the manpage fails to be transcoded (as in iconv with transcode option). Usually, the manpages should be installed in /usr/share/man/ja/ directory, to allow man-db to do the transcoding. UPDATE: After discussion in debian-users@jp: considering the current situations where Fedora Core et al are converting to UTF-8 for manpages, and warning on 'legacy encodings', Debian may need to start considering shipping utf-8 manpages, or have to go through the hassle of having to convert every manpage to the 'legacy encodings'.