On Fri, May 21, 2010 at 10:05:50AM -0700, Bart Smaalders wrote: > On 20/05/2010 18:50, Roland Mainz wrote: > > IMO this case should either allow the use of multibyte characters or > > expcitly refer to bytes/ASCII characters (see below). > > Since there is no way of storing encoding information along with > the username, any attempt to use multibyte will result in display > issues on multi-locale environments; whether or not this is > satisfactory is up to the users in question.
We could have a convention that usernames (and groupnames, and...) must be stored encoded in UTF-8, then make sure all relevant applications and/or library functions (e.g., getXbyYs) convert between the user's locale's codeset and UTF-8. But that's not this case. > The OS should 8 bit clean wrt usernames just as it is for filenames; > the same problems exist there. For filenames we already have the option to prefer UTF-8 in ZFS (well, ZFS can't enforce it in the fact of non-UTF-8 filenames whose encodings make for valid UTF-8 encodings, but you see the point). Eventually we'll have to tackle this problem, I'm sure. And indeed, that's "not this case". Roland's right that there should be a _warning_ to use only ASCII characters in usernames, though that is still not this case (it's at best a documentation bug, assuming such a warning isn't already there). Nico -- _______________________________________________ opensolaris-arc mailing list [email protected]
