get-char-code-property: reports incorrect char names?

get-char-code-property: reports incorrect char names?

[T.V. Raman]

This is using the latest bzr snapshot, with the latest
unicodedata.txt from the Unicode  consortium.


I noticed that I was hearing incorrect character names for
unicode chars from Emacspeak --- and it's possibly due to a bug
in get-char-code-property.

The solution that works  describe-char-unicode-data --- is
declared to be semi-obsolete.

To see the problem, see code fragments and result of evaluation:

The  character used is the "cent sign" ==
¢ inserted via C-x 8c in the buffer.

(loop for p in
      (mapcar 'car char-code-property-alist)
      do
      (insert
       (format "%s: %s\n" p (get-char-code-property ?¢ p ))))

titlecase: nil
lowercase: nil
uppercase: nil
iso-10646-comment: nil
old-name: nil
mirrored: N
numeric-value: nil
digit-value: nil
decimal-digit-value: nil
decomposition: nil
bidi-class: ET
canonical-combining-class: 0
general-category: Sc
name: SUPERSET NYA FA SIGN


Compared to what describe-char-unicode-data returns for the same
character: note particularly the Name field.
(describe-char-unicode-data ?¢)
(("Name" "CENT SIGN")
 ("Category" "Symbol, Currency")
 ("Combining class" "Sc")
 ("Bidi category" "Sc")
 ("Decomposition" nil)
 ("Decimal digit value" nil)
 ("Digit value" nil)
 ("Numeric value" nil)
 ("Mirrored" nil)
 ("Old name" nil)
 ("ISO 10646 comment" nil)
 ("Uppercase" nil)
 ...)

--

Re: get-char-code-property: reports incorrect char names?

[Eli Zaretskii]

> Date: Tue, 31 Aug 2010 17:36:35 -0700
> From: "T.V. Raman" <tv.raman.tv@gmail.com>
> Cc: tv.raman.tv@gmail.com
> 
> This is using the latest bzr snapshot, with the latest
> unicodedata.txt from the Unicode  consortium.

Did you actually mean "from the Unicode  consortium", i.e. you are
using a file that is different from what we have in the repository?

Introduction of a new UnicodeData.txt file requires to run several
commands in the admin subdirectory, and then rebuild several Lisp
files.  Did you do all that?

> I noticed that I was hearing incorrect character names for
> unicode chars from Emacspeak --- and it's possibly due to a bug
> in get-char-code-property.
> 
> The solution that works  describe-char-unicode-data --- is
> declared to be semi-obsolete.
> 
> To see the problem, see code fragments and result of evaluation:
> 
> The  character used is the "cent sign" ==
> ¢ inserted via C-x 8c in the buffer.
> 
> (loop for p in
>       (mapcar 'car char-code-property-alist)
>       do
>       (insert
>        (format "%s: %s\n" p (get-char-code-property ?¢ p ))))
> 
> titlecase: nil
> lowercase: nil
> uppercase: nil
> iso-10646-comment: nil
> old-name: nil
> mirrored: N
> numeric-value: nil
> digit-value: nil
> decimal-digit-value: nil
> decomposition: nil
> bidi-class: ET
> canonical-combining-class: 0
> general-category: Sc
> name: SUPERSET NYA FA SIGN

I cannot reproduce this with today's build of the trunk.  I get "CENT
SIGN" for the `name' property of a character inserted by "C-x 8c".
Also, there's no character named "SUPERSET NYA FA SIGN" in
UnicodeData.txt that's in the repository.

Does this happen for you in "emacs -Q"?  If not, there's some other
factor at work here.

Re: get-char-code-property: reports incorrect char names?

[Stefan Monnier]

> I noticed that I was hearing incorrect character names for
> unicode chars from Emacspeak --- and it's possibly due to a bug
> in get-char-code-property.

Please use M-x report-emacs-bug rather than posting to emacs-devel.
I can't reproduce your problem.  It's probably due to a some
coding-system problem.


        Stefan