NTEmacs, German, ispell

NTEmacs, German, ispell

[Sören Vogel]

Hi,

I'm using ntemacs 21.2.1 on German Windows XP. Spell checking does
ispell and in order to provide German Umlaut support in
conjunction with the new dictionary I placed the following lines
into my .emacs:

 '(current-language-environment "German")
 '(default-input-method "german-postfix");; ???
 '(enable-multibyte-characters nil)
 '(ispell-help-in-bufferp t)
 '(ispell-local-dictionary "deutsch8")
 '(set-language-environment "German")

Seems like emacs recognizes German words, e.g. müssen (must, have
to). But ispell marks the whole word red and suggests _the same
word_ from the spelling context menu! Meaning: "You should change
"müssen" to "müssen". Okay, I cange (click on the suggested
alternative). But the word remains marked red -- misspelled. Only
picking up the word into my private dict ~/.ispell-private makes
the mark dissappearing. The line in the private dict then
represents

mu"ssen

I've found out that some distributed ispell German dicts contain a
bug by representing German Umlaute with "u instead of u". So I
changed the default dict with the sed command

sed 's/"\(.\)/\1"/g' wrong-dictionary > proper-dictionary

but emacs spelling behavior did not change.

Anyone suggestions?

Thanks a lot, Sören from Germany.

PS: Could you please reply via email, too.

Re: NTEmacs, German, ispell

[Eli Zaretskii]

> From: =?ISO-8859-1?Q?S=F6ren_Vogel?= <soeren.vogel@phil.tu-chemnitz.de>
> Newsgroups: gnu.emacs.help
> Date: Sat, 18 Jan 2003 14:53:42 +0100
> 
> But ispell marks the whole word red and suggests _the same
> word_ from the spelling context menu!

I suspect that you type Latin-9 characters while Ispell suggests
Latin-1 characters (they are different characters as far as Emacs is
concerned).

What does Emacs print if you go to the seemingly-identical characters
and type "C-u C-x ="?

Re: NTEmacs, German, ispell

[Sören Vogel]

Hello Eli,

On Sat, 18 Jan 2003, Eli Zaretskii wrote:

  character: s (0163, 115, 0x73)
    charset: ascii (ASCII (ISO646 IRV))
 code point: 115
     syntax: word
   category: a:ASCII   l:Latin
buffer code: 0x73
  file code: 0x73 (encoded by coding system raw-text-dos)
       font:
	     -outline-Courier
New-bold-r-normal-normal-13-97-96-96-c-80-iso8859-1

Doesn't say nothing to me :-/

Sören.

Re: NTEmacs, German, ispell

[Eli Zaretskii]

On Sat, 18 Jan 2003, =?ISO-8859-1?Q?S=F6ren_Vogel?= wrote:

>   character: s (0163, 115, 0x73)
>     charset: ascii (ASCII (ISO646 IRV))

I meant non-ASCII characters, such as u-umlaut.  ASCII characters are 
guaranteed to be identical everywhere in Emacs.

> Doesn't say nothing to me :-/

Well, how about reading the description of the command invoked by
"C-x ="?  Here's a key sequence that should land you at the appropriate 
place in the on-line manual:

	C-h i d m emacs RET i what-cursor-position RET

Does the description there help to understand how would "C-u C-x =" be 
useful in solving your mystery?

Re: NTEmacs, German, ispell

[Sören Vogel]

Hi Eli,

On Sun, 19 Jan 2003, Eli Zaretskii wrote:

> Does the description there help to understand how would "C-u C-x
> =" be useful in solving your mystery?

Great, helped :-)! And here is what emacs says:

  character: ü (0374, 252, 0xfc)
    charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))
 code point: 252
     syntax: word
   category:
buffer code: 0xFC
  file code: 0xFC (encoded by coding system raw-text-dos)
       font:
	     -outline-Courier
New-bold-r-normal-normal-13-97-96-96-c-80-iso8859-

Think this describes the character ü, and now I have to check,
whether emacs uses the same set to display characters. In addition
I expect me to check whether ispell uses the same set?

And how can that be done?

Sören.

Re: NTEmacs, German, ispell

[Kai Großjohann]

Sören Vogel <soeren.vogel@phil.tu-chemnitz.de> writes:

>   character: ü (0374, 252, 0xfc)
>     charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))

eight-bit-graphic is the worst character set of all the possible ones
that can display this glyph.

Does your keyboard generate Latin-1?  So how about doing C-x RET k
latin-1 RET?  And then enter ü and do the C-u C-x = dance on it.  Do
you now get latin-1 instead of eight-bit-graphic?

Does ispell work better on that word?
-- 
Ambibibentists unite!

Re: NTEmacs, German, ispell

[Eli Zaretskii]

On Sun, 19 Jan 2003, =?ISO-8859-1?Q?S=F6ren_Vogel?= wrote:

>   character: ü (0374, 252, 0xfc)
>     charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))

This means Emacs didn't recognized the file as being in Latin-1 encoding, 
probably because it has some Windows-specific characters that Microsoft 
added to the Latin-1 charset.

Does it help to visit the file with "C-x RET c latin-1 RET C-x C-f" 
instead of the usual "C-x C-f"?  That is, if you use C-x RET C, does 
Ispell work as you'd expect?

Re: NTEmacs, German, ispell

[Sören Vogel]

I hope f'ups to more than one person is possible ...?

Kai wrote:

> ... how about doing C-x RET k latin-1 RET?  And then enter ü and
> do the C-u C-x = dance on it. Do you now get latin-1 instead of
> eight-bit-graphic?

Eli wrote:

> Does it help to visit the file with "C-x RET c latin-1 RET C-x
> C-f" instead of the usual "C-x C-f"?

No, remains saying eight-bit-graphic and ispell does not work
properly in both cases: defining the coding system or loading from
within a defined one. Hm, maybe emacs in fact has _no_ problem
with my stuff but ispell does so? And how about the mystery that
ispell suggests exactly the same word "müssen" but does not (or
does -- that't not to observe) replace the "misspelled" but
remains saying: wrong. Perhaps I should try to set up the ultimate
options to work with a German keyboard on WinXP-text-files and try
all stuff again -- but what are the options to place into
".emacs"? Meaning: Does anyone successfully work with emacs on
German WinXP in conjunction with ispell and what did he/she set
up?

Thanks a lot, Sören.

Re: NTEmacs, German, ispell

[Eli Zaretskii]

On Sun, 19 Jan 2003, =?ISO-8859-1?Q?S=F6ren_Vogel?= wrote:

> > Does it help to visit the file with "C-x RET c latin-1 RET C-x
> > C-f" instead of the usual "C-x C-f"?
> 
> No, remains saying eight-bit-graphic and ispell does not work
> properly in both cases: defining the coding system or loading from
> within a defined one.

Strange.  What kind of file is that, anyway?

What happens if you begin with an empty buffer and then type German words 
using those characters--does Ispell work then?

> Hm, maybe emacs in fact has _no_ problem
> with my stuff but ispell does so?

The problem is that Emacs doesn't recognize characters such as u-umlaut 
as Latin-1 characters.  (That's why it says eight-bit-graphic instead.)  
Ispell is set up in your locale for Latin-1 characters, so it cannot work 
with other characters.

> And how about the mystery that
> ispell suggests exactly the same word "müssen" but does not (or
> does -- that't not to observe) replace the "misspelled" but
> remains saying: wrong.

It's not the same word: the character code of what is displayed as 
u-umlaut is different.  The illusion that the words are identical is 
because both characters are (by default) displayed the same.  Try 
accepting Ispell's correction, and then do a "C-u C-x =" on u-umlaut in 
the replaced word, and you will see the difference.

> Perhaps I should try to set up the ultimate
> options to work with a German keyboard on WinXP-text-files

What ``ultimate options'' are you talking about?  What do you have now in 
your .emacs?

Re: NTEmacs, German, ispell

[Heinz Rommerskirchen]

Sören Vogel <soeren.vogel@phil.tu-chemnitz.de> writes:

> Hi,
> 
> I'm using ntemacs 21.2.1 on German Windows XP. Spell checking does
> ispell and in order to provide German Umlaut support in
> conjunction with the new dictionary I placed the following lines
> into my .emacs:
> 
>  '(current-language-environment "German")
>  '(default-input-method "german-postfix");; ???
>  '(enable-multibyte-characters nil)
>  '(ispell-help-in-bufferp t)
>  '(ispell-local-dictionary "deutsch8")
>  '(set-language-environment "German")

I'm no expert but the line 
  (enable-multibyte-characters nil)
looks fishy to me. Also 
  (default-input-method "german-postfix")
is not needed, if you have a German keyboard

-- 
Regards

Heinz

Re: NTEmacs, German, ispell

[Sören Vogel]

Hi All,

I've tried a lot and figured out that nothing works so good as to
download a new binary and link the spell checking to aspell with
options

(setq-default ispell-program-name "aspell")
(setq-default ispell-extra-args '("--lang=de"))

defined in my .emacs.

I've downloaded the new emacs-windows-fullbin.tar.gz (?) from
ftp.gnu.org using FileZilla and now everything works fine, meaning
German Umlaute are recognized to be usual latin-1 characters. I
guess some corrupt char tables were responsible for the bad
behaviour. This might have occured after transferring the
whole emacs-dir from one computer to another via ftp.

Currently I try to incorporate aspell into and the first trials
look good. In addition it is praised to be much better as ispell,
especially for German language.

Special thanks to Eli and Kai for their helping posts.

Sören (with a latin-1-OE) :-).