Benjamin.Riefenstahl at epost.de
Sat Mar 6 18:18:24 GMT 2004
Andrew Bartlett <abartlet at samba.org> writes:
> Actually, we make it work pretty well. The issues we are having
> here are because people keep wanting to stretch the rules!
What are we talking about here? Isn't this just a question of
optimizing things for systems that do conform to your rules? There
are systems that do not conform, so they don't get those extra
optimizations. Is there anything else?
> The rules where:
> - Behaves like a C string (null terminated, no intermediate nulls)
> - no ASCII in subsequent multibyte characters
> - all multibyte characters start with the high bit set.
Just to reiterate: You don't have the second of these with some of the
encodings we discussed, unless you do conversions of some kind. You
don't have the third with de-composed UTF-8. IOW you just don't have
them in some existing Unix filesystems today.
Arguing that it's other people's fault isn't going to change that you
(and more importantly the users of those systems) don't have those
assurances. The users are just victims here, they didn't decide any
of this, and they more often than not don't have a chance decide it.
More information about the samba-technical