Strange charset selection when posting in iso-8859-1

Strange charset selection when posting in iso-8859-1

[Peter Simons]

Hi,

I've run into a problem with Gnus 5.9.0 when posting articles (or
e-mails) that contain ISO-8859-1 characters. When I hit C-c C-c
to send the posting off, Emacs asks for the charset the article
would be using, but the selection it offers doesn't contain
ISO-8859-1! There is "latin-iso-8859-1" (or something like that),
but that's not a valid charset name for RFC articles, so I
essentially can't post non-ASCII articles at all right now. :-(

The problem started to appear after I added

  (setq-default enable-multibyte-characters nil)

to my ~/.emacs file. Now, I don't want multi-byte mode enabled,
but I don't want to forego umlauts or guillemots either.

Does anyone have a recommendation for me how to remedy this
problem?

Peter

Re: Strange charset selection when posting in iso-8859-1

[Reiner Steib]

On Mon, Sep 05 2005, Peter Simons wrote:

> The problem started to appear after I added
>
>   (setq-default enable-multibyte-characters nil)
>
> to my ~/.emacs file. Now, I don't want multi-byte mode enabled,

Why?

> but I don't want to forego umlauts or guillemots either.
>
> Does anyone have a recommendation for me how to remedy this
> problem?

Remove that line.  Gnus doesn't work correctly in unibyte mode, AFAIK.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: Strange charset selection when posting in iso-8859-1

[Peter Simons]

Reiner Steib writes:

 >> The problem started to appear after I added
 >>
 >> (setq-default enable-multibyte-characters nil)
 >>
 >> to my ~/.emacs file. Now, I don't want multi-byte mode
 >> enabled,
 >
 > Why?

Because I had lots of trouble with Emacs choosing weird
encodings for _all_ kind of files when I had it enabled.


 > Gnus doesn't work correctly in unibyte mode, AFAIK.

Well, using the proper charset identifiers as defined in
RFC1521 would fix the problem for me, as far as I can tell.
I don't see why that wouldn't be possible. What is the
rationale behind a choice like "latin1-iso-8859-1"? Isn't
that a pleonasm anyway?

Peter

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>   (setq-default enable-multibyte-characters nil)

If you have problems that seem to be fixed by the above, please use
M-x report-emacs-bug RET to report them.


        Stefan

Re: Strange charset selection when posting in iso-8859-1

[Mark Hood]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
> >   (setq-default enable-multibyte-characters nil)
> 
> If you have problems that seem to be fixed by the above, please use
> M-x report-emacs-bug RET to report them.

Ever since Emacs 20 I have had the following lines in my .emacs:

(setq default-enable-multibyte-characters nil)
(set-language-environment "Latin-1")

Otherwise I was constantly fighting problems with coding systems.  

I often work directly with binary files, but even files I thought were
supposed to be plain unibyte text would often cause Emacs V20 and
later to ask me strange and confusing questions when reading and
writing them.  Much of the large library of Emacs lisp I had developed
for myself was broken due to multibyte issues.

Really, it was much easier to put those two lines in my .emacs then to
try to sort out all those issues in attempts to make reproducible and
understandable bug reports.  My usage of Emacs is always within the
context of European languages, so unibyte mode works very well for me.

I was stuck on Emacs 19.34 for years until I discovered that those two
lines reliably put Emacs V20 and later into unibyte mode.

-- Mark Hood

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>> >   (setq-default enable-multibyte-characters nil)
>> 
>> If you have problems that seem to be fixed by the above, please use
>> M-x report-emacs-bug RET to report them.

> Ever since Emacs 20 I have had the following lines in my .emacs:

> (setq default-enable-multibyte-characters nil)
> (set-language-environment "Latin-1")

> Otherwise I was constantly fighting problems with coding systems.  

Notice the use of the past tense in your above sentence.  Emacs-20
introduced problems with coding-systems, of course, that was unavoidable.
Many years have passed since then: third-party packages have been fixed,
defaults and heuristics have been improved.

Please try it again.

> I was stuck on Emacs 19.34 for years until I discovered that those two
> lines reliably put Emacs V20 and later into unibyte mode.

You could have read the NEWS file (C-h N) which would have told you
the same.


        Stefan

Re: Strange charset selection when posting in iso-8859-1

[Ivan Boldyrev]

On 9224 day of my life Stefan Monnier wrote:
>>   (setq-default enable-multibyte-characters nil)
>
> If you have problems that seem to be fixed by the above, please use
> M-x report-emacs-bug RET to report them.

It is not bug, it is misfeature.  If file contains garbage, Emacs
attempts to select another encoding.  Best way to fix it is to remove
garbage from file or add file cookie (see info://(emacs)File+Variables") :) 

-- 
Ivan Boldyrev

              "Assembly of Japanese bicycle require great peace of mind."

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>>> (setq-default enable-multibyte-characters nil)
>> 
>> If you have problems that seem to be fixed by the above, please use
>> M-x report-emacs-bug RET to report them.

> It is not bug, it is misfeature.

M-x report-emacs-bug is not just for bugs, but for misfeatures and feature
requests as well.


        Stefan

Re: Strange charset selection when posting in iso-8859-1

[Peter Simons]

Stefan Monnier writes:

 >> (setq-default enable-multibyte-characters nil)
 >
 > If you have problems that seem to be fixed by the above,
 > please use M-x report-emacs-bug RET to report them.

Like some others have said too: the problem isn't that Emacs
would have any "bugs" I could report, it's just that I do
_not_ want it to guess any encodings. I use iso-8859-1, and
that's good enough for me.

Now, I still wonder why Gnus offers latin1-iso-8859-1
instead of the correct name. If this would be fixed, I'd be
a happy camper.

Peter

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>>> (setq-default enable-multibyte-characters nil)
>> 
>> If you have problems that seem to be fixed by the above,
>> please use M-x report-emacs-bug RET to report them.

> Like some others have said too: the problem isn't that Emacs
> would have any "bugs" I could report, it's just that I do
> _not_ want it to guess any encodings.  I use iso-8859-1, and
> that's good enough for me.

Suit yourself.  As the multibyte support improves and beomces more
sophisticated, it's becoming increasingly difficult to make sure that both
the multibyte and the unibyte modes work correctly, and the multibyte mode
is the most commonly used, so unless you remove the above line from your
Emacs, you're likely to bump into an increasing number of annoyances.

Just like you, the only languages I know are all covered by latin-1, and
believe me, we're not the only two.  Most of those people use the default
multibyte mode without any problem.  Try it.


        Stefan

Re: Strange charset selection when posting in iso-8859-1

[Peter Simons]

Stefan Monnier writes:

 > [...] unless you remove the above line from your Emacs,
 > you're likely to bump into an increasing number of
 > annoyances.

The number of annoyances I bumped into with multi-byte mode
enabled was much higher than without it. Others may believe that
wouldn't be true, but I reckon my word is more authoritative than
other people's word when it comes to what annoys _me_.

Now, I don't recall whether I mentioned it already, but there is
a curious problem in Gnus, namely the fact that it uses the name
"latin1-iso-8859-1" to designate the encoding the rest of the
world calls "iso-8859-1". I realize that fixing this problem
requires re-writing Gnus from scratch, more or less, but it is
not going to become less of an effort by discussing a completely
different topic for the next two weeks.

It would be just GREAT if someone with the power to make changes
happen could comment on that subject.

Peter

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>> [...] unless you remove the above line from your Emacs,
>> you're likely to bump into an increasing number of
>> annoyances.

> The number of annoyances I bumped into with multi-byte mode
> enabled was much higher than without it.

I believe you.  That's why the unibyte mode was introduced: because the
multibyte mode was not robust enough yet.  We're getting to the point where
the multibyte mode has fewer problems than the unibyte mode.

> Now, I don't recall whether I mentioned it already, but there is
> a curious problem in Gnus, namely the fact that it uses the name
> "latin1-iso-8859-1" to designate the encoding the rest of the
> world calls "iso-8859-1". I realize that fixing this problem
> requires re-writing Gnus from scratch, more or less, but it is
> not going to become less of an effort by discussing a completely
> different topic for the next two weeks.

Why not post a bug report?


        Stefan

Re: Strange charset selection when posting in iso-8859-1

[Peter Simons]

Stefan Monnier writes:

 >> Gnus [uses] "latin1-iso-8859-1" to designate the
 >> encoding the rest of the world calls "iso-8859-1".
 >
 > Why not post a bug report?

I thought I _was_ reporting the problem here. 

Anyway, I fixed it by now, so more reporting won't be
necessary. Thank you for your help.

Peter

Re: Strange charset selection when posting in iso-8859-1

[Stefan Monnier]

>>> Gnus [uses] "latin1-iso-8859-1" to designate the
>>> encoding the rest of the world calls "iso-8859-1".
>> Why not post a bug report?
> I thought I _was_ reporting the problem here.

This newsgroup is to get help about Gnus.  Some Gnus maintainers may read
it, but not all of them and not always.
To report bugs, use M-x gnus-bug RET, which will send an email to
(all) the maintainers.


        Stefan