Re: File name completion with non-ascii filenames

[Kenichi Handa]

"Eli Zaretskii" <address@hidden> writes:
>>  From: Richard Stallman <address@hidden>
>>  Date: Tue, 18 Dec 2001 22:59:50 -0700 (MST)
>>  
>>  + static int
>>  + scmp (s1, s2, len)
>>  +      register unsigned char *s1, *s2;
>>  +      int len;
>>  + {
>>  +   register int l = len;
>>  + 
>>  +   if (completion_ignore_case)
>>  +     {
>>  +       while (l && DOWNCASE (*s1++) == DOWNCASE (*s2++))
>>  +   l--;
>>  +     }

> Is this the proper way of comparing unibyte characters
> case-insensitively?  That is, are the tables used by DOWNCASE set up
> for unibyte characters as well?

> Come to think about this, what does completion_ignore_case have to do
> with comparing file names?  I think it's not relevant, so I think we
> should remove the case-insensitive branch.

The case-table is setup correctly for 8-bit characters in
each language environment.  But, FILE arg of
file_name_completion may contain multibyte characters that
can't be encoded by file-name-coding-system.  In that case,
ENCODE_FILE produces inproper unibyte string that will match
with a filename incorrectly.

The general strategy of Emacs for comparing unibyte and
multibyte string is to convert or decode unibyte string to
multibyte.

So, in file_name_completion also, instead of encoding FILE,
we should decode each filenames in the directory by
file-name-coding-system, and use Fstring_compare.

---
Ken'ichi HANDA
address@hidden