Re: Fontsets and ispell

[Kenichi Handa]

In article <address@hidden>, Peter Dyballa <address@hidden> writes:

> Hello!
> I tried to spell-check a LaTeX file in an AUCTeX (version: 11.55) 
> buffer. When ispell hit the word sérifs (of sans-sérifs) this error was 
> reported:

>       Debugger entered--Lisp error: (error "Ispell misalignment: word 
> `s\x8e9rifs' point 1574; probably incompatible versions")
[...]
> So obviously the é was passed as something else but itself. C-u C-x = 
> reports this:

>         character: é (07551, 3945, 0xf69, U+00E9)
>           charset: latin-iso8859-15 (Right-Hand Part of Latin Alphabet 9 
> (ISO/IEC 8859-15): ISO-IR-203.)
[...]
> Although it's stated 'buffer code: 0x8E 0xE9' the buffer's modeline 
> starts with "-0:--", i.e. ISO Latin-15. ispell is set up to work like 
> this customization:

>   '(ispell-local-dictionary-alist (quote (("english" "[a-zA-Z]" 
> "[^a-zA-Z]" "[']" t ("-C") "~tex" iso-8859-1) (nil 
> "[a-zA-ZÄÖÜßäöüÇÀÂÁÉÊÈÍÎÌÓÔÒÚÛÙÑÆçàâáéêèíîìóôòúûùñæ]" 
> "[^a-zA-ZÄÖÜßäöüÇÀÂÁÉÊÈÍÎÌÓÔÒÚÛÙÑÆçàâáéêèíîìóôòúûùñæ]" "" t ("-d" 
> "german") "~latin1" iso-8859-1))))

Emacs distinguishes latin-1 and latin-15 characters.

Please specify CASECHARS and NOT-CASECHARS in unibyte
(i.e. using octal form of byte sequence for iso-8859-1)
without using '-' between eight-bit chars (see
lisp/textmodes/ispell.el.  Then the latest Emacs generates a
proper multibyte string that covers all equivalent
characters in CASECHARS and NOT-CASECHARS.

I've just installed this change in the docstring of
ispell-dictionary-alist, but it seems the English is not
good.   Could someone please improve it?

*** ispell.el   04 Feb 2005 09:28:49 +0900      1.156
--- ispell.el   21 Feb 2005 11:13:16 +0900      
***************
*** 671,676 ****
--- 671,682 ----
  If you want OTHERCHARS to be empty, use the empty string.
  Hint: regexp syntax requires the hyphen to be declared first here.
  
+ CASECHAS, NOT-CASECHARS, and OTHERCHARS must be a unibyte string
+ containing bytes of CHARACTER-SET.  In addition, if they contain
+ a non-ASCII byte, the regular expression must be a single
+ `character set' construct that doesn't specify a character range
+ for non-ASCII bytes.
+ 
  MANY-OTHERCHARS-P is non-nil when multiple OTHERCHARS are allowed in a word.
  Otherwise only a single OTHERCHARS character is allowed to be part of any
  single word.

---
Ken'ichi HANDA
address@hidden