Re: find-coding-systems inconsistencies

[Kenichi Handa]

In article <address@hidden>, Jesper Harder <address@hidden> writes:

> Consider the following three similar ways of asking the same question:
> 1. (find-coding-systems-for-charsets (list (char-charset ?)))
>    => (raw-text emacs-mule)

This is wrong.  find-coding-systems-for-charsets was written
before supporting utf-X and should be fixed now.

> 2. (find-coding-systems-string "")
>    => (undecided)

> 3. (with-temp-buffer
>      (insert "")
>      (find-coding-systems-region (point-min) (point-max)))

2 and 3 are diffrerent operations.

2a. (find-coding-systems-region (string-make-multibyte "\235"))

3a. (with-temp-buffer
       (insert "\235")
       (set-buffer-multibyte nil)
      (find-coding-systems-region (point-min) (point-max)))

2 and 3a, 3 and 2a are the same operations respectively.

And, the result of 3 (and 2a) depends on the language
environment.  For instance, in Bulgarian lang. env., the
result contain windows-1251 and etc.  That is because
(string-make-multibyte "\235") returns a string that
contains a Cyrillic character in the charset
mule-unibyte-0100-24ff.

> And is \235 really encodable by utf-8?

Yes, because Unicode explicitly contains C1-control
characters.

---
Ken'ichi HANDA
address@hidden