From: David Starner (dvdeug@ispwest.com)
Date: Mon Apr 28 2003 - 12:20:35 EDT
On Mon, Apr 28, 2003 at 08:27:36AM -0700, Jane Liu wrote:
> neither Microsoft Windows nor those popular UNIX
> systems (AIX, Solaris, HP-UX) currently supply the explicit support
> of Unicode normalization at the encoding/converison level
[...]
> If that's true, can we conclude that in order to maintain the
> transperancy and round-trip safty between application and OS, the
> application should not use normalization?
On Posix systems, it's wrong to even treat filenames as Unicode strings;
according to the standard, they are null-terminated byte strings that
can't include 0x2F. From that perspective, filenames are a unique type
that musn't be mangled by any Unicode operation, including conversion
between UTF-8 and UTF-16. Of course, users expect to see them as
strings; one solution for this that also works for the normalization
case is to keep a table of filenames and the Unicode version thereof,
which may be normalized.
-- David Starner - dvdeug@email.ro Ic sæt me on anum leahtrice, ða com heo and bát me!
This archive was generated by hypermail 2.1.5 : Mon Apr 28 2003 - 13:53:54 EDT