On Tue, Nov 28, 2000 at 01:46:00PM -0800, Gisle Aas wrote: > Jarkko Hietaniemi <jhi@iki.fi> writes: > > > Hmmm. I would expect UTF-8-ness (or UTF-16-ness, for that matter) or > > filenames to be a per-filesystem(-mount) property, > > Or perhaps even a per directory property or something to be specified > by locales. Different users on a Linux box might operate with <shudder> :-) Yes, I guess something like LC_FILENAME_ENCODING would be needed. (And as i18n spreads, LC_USERNAME_ENCODING, LC_HOSTNAME_ENCODING, etc) (I just developed a genuine case of VMS envy: having tables, as opposed to one table, as your environment, would be so much nicer. Alas.) > different file name policies on the same file system. One user might A very good point. > use UTF8 filenames, another latin1 and a third some other 8 bit > encoding. > > File names on Unix is really just binary data (with '\0' and '/' Yup, my Latin-1 was a gross assumption. It might be e.g. HP Roman 8, or SJIS, or whatever. > having special meaning). It is up the the applications to impose > character semantics on them if they want. Usually, they luckily don't (except to search for ASCII-safe separators like '.'). > Regards, > Gisle -- $jhi++; # http://www.iki.fi/jhi/ # There is this special biologist word we use for 'stable'. # It is 'dead'. -- Jack CohenThread Previous | Thread Next