Re: UTF-8 breakage

[Reiner Steib <4uce.02.r.steib@gmx.net>]

On Mon, Dec 02 2002, Simon Josefsson wrote:

> Reiner Steib <4uce.02.r.steib@gmx.net> writes:
>> On Sun, Dec 01 2002, Simon Josefsson wrote:

>> I didn't mention (and test) Gnus 5.8.8, because I though that it's not
>> recommended to use Emacs 21.x together with Gnus 5.8.8 (as Gnus 5.9 is
>> shipped with Emacs 21.x).
>
> Ah.  I was thinking of XEmacs users, but perhaps the page was about
> Emacs only.

It is about Emacs only. For XEmacs, I only know that there's the
latin-unity package. But, as I mentioned in the last post, people
report that they can't get proper unification with XEmacs 21.4 and
Gnus 5.8.8.

>>> I think Oort is supposed to unify at least across ISO-8859-*
>>> automatically when possible [...] räksmörgås r€ksm€rg€s

>> When I *dis*able unify-8859-on-{de,en}coding-mode and add a `ß'
>> (Latin-1) replying to your message, I get utf-8 (checked with
>> `mml-preview') instead of Latin-9.
>
> Do you instruct Emacs or Gnus to prefer UTF-8 anywhere?  

No, at least not by purpose. I tested it with a minimal setup now:

- Starting Emacs 21.1 (LC_CTYPE=en_US.ISO_8859-1)
- reloading `utf-8', enable unify-8859-on-{de,en}coding-mode [0]
- start Oort Gnus
- replying to your message and inserting Latin-1 characters
- disable unify-8859-on-{de,en}coding-mode
- C-u C-c RET P
==> utf-8

- Starting Emacs 21.1 and Oort Gnus (LC_CTYPE=en_US.ISO_8859-1)
- replying to your message and inserting Latin-1 characters
  (LC_CTYPE=en_US.ISO_8859-1)
- C-u C-c RET P
==> Latin-9

> I think the unification is to the preferred charset.

It works (without unify-8-o-*-mode) sometimes, but not always:

- Starting Emacs 21.1 and Oort Gnus (LC_CTYPE=en_US.ISO_8859-1)
- Reply to news:m3fzthfiyl.fsf@merlin.emma.line.org (Article with a
  `ß' from Latin-9).
- insert a Latin-1 Currency sign or acute accent[1] (not codable in
  Latin-9!)
- C-u C-c RET P
==> "A message part needs to be split into 2 charset parts.  Really send?"

With unify-8-o-*-mode, I get unification to Latin-1.

- Starting Emacs 21.1 and Oort Gnus (LC_CTYPE=en_US.ISO_8859-1)
- start with a new article
- insert a Latin-1 `ä'
- insert a unicode € [2]
- C-u C-c RET P
==> "... split into 2 charset parts ..."
    --> charset=iso-8859-15 (Part 1)
    --> charset=charset=utf-8 (Part 2)

With unify-8-o-*-mode, I get unification to Latin-9 in this case.

So I think that even with Oort ucs-tables is still useful.

Bye, Reiner.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[0] This is what the file `rs-gnus-unify.el' does (apart from some tests).
    <URL: http://my.gnus.org/Members/rsteib/howto_unify/>
    
[1] (insert (string (make-char 'latin-iso8859-1 164))) ; CURRENCY SIGN
    (insert (string (make-char 'latin-iso8859-1 180))) ; ACUTE ACCENT

[2] (insert (string (make-char 'mule-unicode-0100-24ff 116 76))); EURO
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo--- PGP key available via WWW   http://rsteib.home.pages.de/