Re: report on [bad] status of i18n of gnome apps - somebody shouldexplicitly care about it

[Christian Rose <menthos bigfoot com>]

Pablo Saratxaga wrote:
> IMHO there is no solution to that problem (on Unix) and the reason is simply
> the file systems used don't have any info on charset used.
> On Windows on the other side unicode is used for filenames, always, so if
> a different encoding is used by programs then the standard libraries would
> transparently reencode the names.
> 
> On a Linux system the equivalent would be to always use utf-8 to store
> physycally on the ext2 partitions, and have the GNU libc modified so
> that functions dealing with file names transpartently call iconv() to
> convert to/from the user encoding (if the user is not in utf-8).
> 
> That way utf-8 enabled programs could use utf-8 names for files; and
> non utf-8 aware programs will simply have the libc convert the names for them.
> 
> Handling that at the filesystem+libc level is imho the only way to do it
> properly, and keep old programs working.
> Implementing that at any other level will make all programs not using the
> specific lib implementing it to display wrong file names (and conversly,
> programs using that lib will wrongly display file names created by
> programs not using the lib).
> Implementing it at the libc level has the advantages that all programs use
> the libc (and statically linked ones are not aimed at i18n usage).

Hmm, is this request known to glibc maintainers? Else I'd also like to
see a request for this added... :)

Of course, the glibc maintainers might just say that all locales should
use utf-8 in the future...


Christian