Re: Q: something like autoload for coding-systems?

[Kenichi Handa]

Eli Zaretskii <address@hidden> writes:
> On Tue, 6 Nov 2001, Richard Stallman wrote:
>>  Is it possible to identify just a part of the coding system info
>>  which is sufficient for find-coding-systems-region, have that part
>>  always loaded, and autoload the rest?

It's possible but not worth doing that because, then, the
rest part is very small as I wrote below.

> It seems like all you need is for char-coding-system-table to know
> about all the possible coding systems.  Handa-san, is that right?

Yes.  And if we setup all information in
char-coding-system-table, loading of coding systems is
almost finished.  The biggest part of coding system is a
mapping table, and without mapping table, we can't have a
complete char-coding-system-table.

By the way, when we shift to Unicode, the situation gets
more difficult, because we must have lots of mapping table
for such legacy charset as latin-X, JIS, GB, KCS, etc.  (As
I wrote before, in unicode base Emacs, mapping table should
be for charsets, not for coding-systems).  Anyway, I think a
practical solution will be to give up checking all possible
coding systems in find-coding-systems-region, but checking
only these:
        o buffer-file-coding-system
        o several commonly used ones
        o what specified by the current language environment
        o what a user requested

---
Ken'ichi HANDA
address@hidden