Look ma: RFÇ2047

Look ma: RFÇ2047

[Lars Magne Ingebrigtsen]

I've now slashed away at the charset thingies, and there's now
basically just one thing that controls encoding when posting --
gnus-group-posting-charset-alist -- and one that says what the default 
charset is when reading -- gnus-group-charset-alist.

It's tempting to use a value such as:

(defcustom gnus-group-charset-alist 
  '(("^hk\\>\\|^tw\\>\\|\\<big5\\>" cn-big5)
    ("^cn\\>\\|\\<chinese\\>" cn-gb-2312)
    ("^fj\\>\\|^japan\\>" iso-2022-jp-2)
    ("^relcom\\>" koi8-r)
    (".*" iso-8859-1)))

That is, default all "unknown" hierarchies to iso-8859-1, but, er, I
don't know...

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Look ma: RFÇ2047

[Russ Allbery]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> It's tempting to use a value such as:

> (defcustom gnus-group-charset-alist 
>   '(("^hk\\>\\|^tw\\>\\|\\<big5\\>" cn-big5)
>     ("^cn\\>\\|\\<chinese\\>" cn-gb-2312)
>     ("^fj\\>\\|^japan\\>" iso-2022-jp-2)
>     ("^relcom\\>" koi8-r)
>     (".*" iso-8859-1)))

> That is, default all "unknown" hierarchies to iso-8859-1, but, er, I
> don't know...

I think that's more likely to be right than most other choices.  Not sure
what else you would default it to?  The only other obvious choice in the
current environment would be something like us-ascii, and since iso-8859-1
is a superset....

-- 
Russ Allbery (rra@stanford.edu)         <URL:http://www.eyrie.org/~eagle/>

Re: Look ma: RFÇ2047

[Lars Magne Ingebrigtsen]

Russ Allbery <rra@stanford.edu> writes:

> > That is, default all "unknown" hierarchies to iso-8859-1, but, er, I
> > don't know...
> 
> I think that's more likely to be right than most other choices.  Not sure
> what else you would default it to?  The only other obvious choice in the
> current environment would be something like us-ascii, and since iso-8859-1
> is a superset....

If we default to us-ascii, then 8-bit characters will be displayed as
\245.  This will be wrong, but at least it will be equally wrong for
everybody on the entire planet.  :-)

But even though this has a certain jennesequa, I think it's more
practical to default to iso-8859-1, since this is the solution that
will be correct for most people, since most people who post without a
charset parameter are people who post using iso-8859-1, recent
psychic statistics have shown.

So I'll default to iso-8859-1 when reading.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Look ma: RFÇ2047

[Kai.Grossjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

  > But even though this has a certain jennesequa, I think it's more
  > practical to default to iso-8859-1, since this is the solution that
  > will be correct for most people, since most people who post without a
  > charset parameter are people who post using iso-8859-1, recent
  > psychic statistics have shown.

Isn't it obvious that you should connect to the author's psyche to
determine the charset that was meant by the author?  Though maybe you
might also need to connect to the Microserf psyche to really know
what's going on...

I think this would also open up interesting possibilities: how about
using the above mechanism to figure out what the author really meant
rather than just what charset they meant?  Would be easier, too: no
need to connect to a second psyche!

kai
-- 
Abort this operation?   [Abort]  [Cancel]

Re: Look ma: RFÇ2047

[Kai.Grossjohann]

Beware of the missing ":-)" !
kai
-- 
Abort this operation?   [Abort]  [Cancel]

Re: Look ma: RFÇ2047

[Lars Balker Rasmussen]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:
> Beware of the missing ":-)" !

Which?  I didn't see one.
-- 
Lars Balker Rasmussen                                        "Woo hoo!?"
.

Look ma: RFÇ2047

[Hallvard B Furuseth]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> (defcustom gnus-group-charset-alist 
>   '(("^hk\\>\\|^tw\\>\\|\\<big5\\>" cn-big5)
>     ("^cn\\>\\|\\<chinese\\>" cn-gb-2312)
>     ("^fj\\>\\|^japan\\>" iso-2022-jp-2)
>     ("^relcom\\>" koi8-r)

Let's see...
      ("^\\(cz\\|hun\\|pl\\|sk\\)\\>" iso-8859-2)
      ("^israel\\>" ???)
      ("^su\\>"     ???)


>     (".*" iso-8859-1)))

Aaargh!

I know Emacs20/MULE has joined "I know better than you what you want"
Macintosh, but Gnus doesn't *have* to follow their example, you know.

Use ("^\\(comp\\|rec\\|alt\\|......\\)\\>" iso-8859-1) instead,
and current-language-environment or whatever for "unkown groups".
It seems more likely that an "unknown group" which the user is
interested in will be a local group with the user's normal charset
than a latin-1 group.
Except international mailinglist groups, of course...


BTW, I don't know if gnus-group-post-charset-alist is a good thing,
except to choose between several possible encodings of a multibyte
message.  If a user usually posts on his local "latin-2 newsgroups" but
once in a while post to a "latin-1 newsgroup", messages to the latter
will most likely contain latin-2 characters.  Instead we need a
recommended way for the user to tell Emacs to switch character set
(font) - which Gnus can notice.

-- 
Hallvard

Re: Look ma: RFÇ2047

[Russ Allbery]

Hallvard B Furuseth <h.b.furuseth@usit.uio.no> writes:

> Let's see...
>       ("^\\(cz\\|hun\\|pl\\|sk\\)\\>" iso-8859-2)
>       ("^israel\\>" ???)
>       ("^su\\>"     ???)

su.* is the internal news hierarchy for Stanford University and is
completely unrelated to the Soviet Union (as correctly indicated in the
canonical control.ctl file, the INN distribution, and the list of public
news hierarchies).

This hasn't stopped people occasionally trying to use it as a hierarchy
for talking about the Soviet Union, but it's still wrong.  :)

-- 
Russ Allbery (rra@stanford.edu)         <URL:http://www.eyrie.org/~eagle/>

Re: Look ma: =?UNKNOWN?Q?RF=C72047?=

[Chris Tessone]

>>>>> "Russ" == Russ Allbery <rra@stanford.edu> writes:

    Russ> This hasn't stopped people occasionally trying to use it as
    Russ> a hierarchy for talking about the Soviet Union, but it's
    Russ> still wrong.  :)

Yup.

There are actually plans, mostly stemming from the denial of a
sci.lang.russian group in a recent referendum, to make a hierarchy for 
Russian newsgroups though. I *think* the plan was for rus.* or
something, but I've forgotten.

-- 
Chris Tessone              tessone@imsa.edu              tessone@fnal.gov
IMSA PH/Postmaster SNPC                     http://www.imsa.edu/~tessone/
print pack"C*",split/\D+/,`echo "16iII*o\U@{$/=$z;[(pop,pop,unpack"H*",<>
)]}\EsMsKsN0[lN*1lK[d2%Sa2/d0<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<J]dsJxp"|dc`

Re: Look ma: RFÇ2047

[Yair Friedman]

Hallvard B Furuseth <h.b.furuseth@usit.uio.no> writes:
> Let's see...
>       ("^israel\\>" ???)

iso-8859-1 until we have proper support for  iso-8859-8
--
Yair.

Re: Look ma: RFÇ2047

[Lars Magne Ingebrigtsen]

Hallvard B Furuseth <h.b.furuseth@usit.uio.no> writes:

> Let's see...
>       ("^\\(cz\\|hun\\|pl\\|sk\\)\\>" iso-8859-2)
>       ("^israel\\>" ???)

Thanks; I've now added these.

> I know Emacs20/MULE has joined "I know better than you what you want"
> Macintosh, but Gnus doesn't *have* to follow their example, you know.

Well, this is only used when reading articles, so getting this wrong
is no catastrophy.

> Use ("^\\(comp\\|rec\\|alt\\|......\\)\\>" iso-8859-1) instead,
> and current-language-environment or whatever for "unkown groups".
> It seems more likely that an "unknown group" which the user is
> interested in will be a local group with the user's normal charset
> than a latin-1 group.

Perhaps.  How does one go from `current-language-environment' to
coding system?

> BTW, I don't know if gnus-group-post-charset-alist is a good thing,
> except to choose between several possible encodings of a multibyte
> message.  If a user usually posts on his local "latin-2 newsgroups" but
> once in a while post to a "latin-1 newsgroup", messages to the latter
> will most likely contain latin-2 characters.  Instead we need a
> recommended way for the user to tell Emacs to switch character set
> (font) - which Gnus can notice.

This variable is only used in non-Mule Emacsen, so this shouldn't occur.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Look ma: RFÇ2047

[Hallvard B Furuseth]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

>> Macintosh, but Gnus doesn't *have* to follow their example, you know.
> 
> Well, this is only used when reading articles, so getting this wrong
> is no catastrophy.

Sure, but not getting it wrong is still better...  If I have already
_told_ Emacs which charset I'm using, why should Gnus override that?

> How does one go from `current-language-environment' to coding system?

I think one asks on gnu.emacs.help :-)

>> BTW, I don't know if gnus-group-post-charset-alist is a good thing,
>> except to choose between several possible encodings of a multibyte
>> message.  If a user usually posts on his local "latin-2 newsgroups" but
>> once in a while post to a "latin-1 newsgroup", messages to the latter
>> will most likely contain latin-2 characters.

This point is independent of whether or not Mule is in use, so...

>> Instead we need a
>> recommended way for the user to tell Emacs to switch character set
>> (font) - which Gnus can notice.
>
> This variable is only used in non-Mule Emacsen, so this shouldn't occur.

that's what I mean: Even non-Mule Emacsen need that recommended way to
switch character set.  Or at least to tell Gnus which character set is
in use.  On non-Mule, that'd be something with set-default-font & co.

-- 
Hallvard

Re: Look ma: RFÇ2047

[Lars Magne Ingebrigtsen]

Hallvard B Furuseth <h.b.furuseth@usit.uio.no> writes:

> Sure, but not getting it wrong is still better...  If I have already
> _told_ Emacs which charset I'm using, why should Gnus override that?

Gnus shouldn't, and won't, once someone tells me how to do that.  :-)

> >> Instead we need a
> >> recommended way for the user to tell Emacs to switch character set
> >> (font) - which Gnus can notice.
> >
> > This variable is only used in non-Mule Emacsen, so this shouldn't occur.
> 
> that's what I mean: Even non-Mule Emacsen need that recommended way to
> switch character set.

The assumption when using a non-Mule Emacs is that you're going to
edit text using only a single charset.  

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen