10:29 a.m.
My ~/.Xmodmap contains:
keysym e = e NoSymbol EuroSign NoSymbol
keysym c = c NoSymbol 0x010000a2 currency
Using that, I can input '€' (U+020AC EURO SIGN),
'¢' (U+00A2 CENT SIGN), and '¤' (U+00A4 CURRENCY SIGN) by typing
the following key sequences:
AltGr e → €
AltGr c → ¢
Shift+AltGr c → ¤
That works fine in all applications *except* XEmacs.
When I use that key sequence to type an € in XEmacs, the cursor
advances but I don't see a glyph. When positioning the cursor on that
character and checking with 'C-u C-x =' I get:
€Char: € (U+20AC greek-iso8859-7 36) point=1032 of 1128(91%) column 0
i.e. the € has been input correctly, the problem is only that it is
represented in the Greek character set, therefore a Greek font is used
and that doesn't have the € character (X11 core fonts are used, not
Xft).
If I input the € by other means, for example
- pasting from another application
- using the input methods t-latn-pre or t-rfc1345 from scim-m17n
I get
Char: € (U+20AC latin-iso8859-15 36) point=1200 of 1660(72%) column 9
and the € is visible.
Why is the Greek charset used when € is input via ~/.Xmodmap?
When trying to input ¢ via ~/.Xmodmap, it is even worse, I only get
the regular ASCII 'c' than.
But ¢ can be input just fine by pasting from another application and
by using an input method like scim-m17n.
So why doesn't this work with ~/.Xmodmap?
--
Mike FABIAN <mfabian(a)suse.de> http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。
I � Unicode
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:16 p.m.
Mike FABIAN writes:
When I use that key sequence to type an € in XEmacs, the cursor
advances but I don't see a glyph.
I suppose you wouldn't, as undoubtedly the Greek font hasn't been
updated since Euclid retired. The font may be broken, too (ie, have
an blank glyph there).
When positioning the cursor on that
character and checking with 'C-u C-x =' I get:
€Char: € (U+20AC greek-iso8859-7 36) point=1032 of 1128(91%) column 0
This is undoubtedly due to the fact that ISO 8859/7 is the
lowest-numbered of the ISO 8859 family to encode the Euro sign. I
suppose that the Unicode-to-Mule algorithm searches the loaded
charsets in some order.
I can't get to this for some time.
Aidan, have any ideas offhand? Do you think stuffing ISO 8859/15 in
the search order immediately after ISO 8859/1 would be a good idea and
easy to do?
Why is the Greek charset used when € is input via ~/.Xmodmap?
Some charset has to be used, and XK_CURRENCY is not a charset
available in Mule.
When trying to input ¢ via ~/.Xmodmap, it is even worse, I only get
the regular ASCII 'c' than.
So why doesn't this work with ~/.Xmodmap?
Violate the standard, get what you deserve. The X11 standard for
Unicode keysyms explicitly says that they may not be used for keys
that already have keysyms.
This also applies to EURO SIGN, which has had an X11 named keysym
(XK_eurosign, 0x20AC) for many years.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:16 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> さんは書きました:
Mike FABIAN writes:
> When I use that key sequence to type an € in XEmacs, the cursor
> advances but I don't see a glyph.
I suppose you wouldn't, as undoubtedly the Greek font hasn't been
updated since Euclid retired. The font may be broken, too (ie, have
an blank glyph there).
The font is the bitmap font
-etl-fixed-medium-r-normal--16-160-72-72-c-80-iso8859-
and it has a blank glyph at 0xa4 indeed.
--
Mike FABIAN <mfabian(a)suse.de> http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。
I � Unicode
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:30 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> さんは書きました:
> When trying to input ¢ via ~/.Xmodmap, it is even worse, I only
get
> the regular ASCII 'c' than.
> So why doesn't this work with ~/.Xmodmap?
Violate the standard, get what you deserve. The X11 standard for
Unicode keysyms explicitly says that they may not be used for keys
that already have keysyms.
This also applies to EURO SIGN, which has had an X11 named keysym
(XK_eurosign, 0x20AC) for many years.
I think it is XK_EuroSign (it is case sensitive, I used that correctly
in my ~/.Xmodmap).
You are right that changing
keysym c = c NoSymbol 0x010000a2 currency
to
keysym c = c NoSymbol cent currency
makes the '¢' work. Great, thank you very much!
--
Mike FABIAN <mfabian(a)suse.de> http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。
I � Unicode
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:32 p.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> さんは書きました:
> €Char: € (U+20AC greek-iso8859-7 36) point=1032 of 1128(91%)
column 0
This is undoubtedly due to the fact that ISO 8859/7 is the
lowest-numbered of the ISO 8859 family to encode the Euro sign. I
suppose that the Unicode-to-Mule algorithm searches the loaded
charsets in some order.
I can't get to this for some time.
Aidan, have any ideas offhand? Do you think stuffing ISO 8859/15 in
the search order immediately after ISO 8859/1 would be a good idea and
easy to do?
> Why is the Greek charset used when € is input via ~/.Xmodmap?
Some charset has to be used, and XK_CURRENCY is not a charset
available in Mule.
But why is there a difference between the input via ~/.Xmodmap, cut &
paste, and input methods like scim-m17n?
When doing cut & paste, probably the target UTF8_STRING is used which
might have an influence to make it work.
But what is the difference between an XIM input method
and ~/.Xmodmap?
--
Mike FABIAN <mfabian(a)suse.de> http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。
I � Unicode
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:48 p.m.
Ar an tríú lá is fiche de mí Meitheamh, scríobh Stephen J. Turnbull:
Mike FABIAN writes:
> When I use that key sequence to type an € in XEmacs, the cursor
> advances but I don't see a glyph.
I suppose you wouldn't, as undoubtedly the Greek font hasn't been
updated since Euclid retired. The font may be broken, too (ie, have
an blank glyph there).
The source of the problem is version incompatibilities with ISO
8859-7. We’ve updated to the 2003 version in 21.5, as has XFree86 and X.org,
but there are still lots of deployed -iso8859-7 fonts without the three
extra code points assigned:
0xA4 0x20AC # EURO SIGN
0xA5 0x20AF # DRACHMA SIGN
0xAA 0x037A # GREEK YPOGEGRAMMENI
The reason the problem doesn’t appear when Mike pastes text from elsewhere
is that the X11 code does the equivalent of (set-character-of-keysym
'EuroSign ?\u20AC) _before_ ~/.xemacs/init.el has been loaded, at which
point (unicode-precedence-list) reflects its value at dump time, and the
converted character reflects that.
Mike, to avoid this, you can safely add
(set-character-of-keysym 'EuroSign ?\u20AC)
to the end of your ~/.xemacs/init.el . The correct fix, though, is to update
the ISO 8859-7 core fonts as used by the X server.
> When positioning the cursor on that
> character and checking with 'C-u C-x =' I get:
>
> €Char: € (U+20AC greek-iso8859-7 36) point=1032 of 1128(91%) column 0
This is undoubtedly due to the fact that ISO 8859/7 is the
lowest-numbered of the ISO 8859 family to encode the Euro sign. I
suppose that the Unicode-to-Mule algorithm searches the loaded
charsets in some order.
I can't get to this for some time.
Aidan, have any ideas offhand? Do you think stuffing ISO 8859/15 in
the search order immediately after ISO 8859/1 would be a good idea and
easy to do?
Easy to do, and a vaguely good idea, but everyone will eventually update
their 8859-7 fonts anyway.
> Why is the Greek charset used when € is input via ~/.Xmodmap?
Some charset has to be used, and XK_CURRENCY is not a charset
available in Mule.
> When trying to input ¢ via ~/.Xmodmap, it is even worse, I only get
> the regular ASCII 'c' than.
That surprises me; your example works for me locally. When you type AltGr +
c and immediately F1 l , does it list U00A2 as the keysym? If so, what does
(get-character-of-keysym 'U00A2) return? (Both in terms of what it looks
like and what C-x = gives you.)
> So why doesn't this work with ~/.Xmodmap?
Violate the standard, get what you deserve. The X11 standard for
Unicode keysyms explicitly says that they may not be used for keys
that already have keysyms.
We accept this behaviour in 21.5, btw, even for ASCII keysyms. And the
keysym EuroSign *is* XK_eurosign.
--
On the quay of the little Black Sea port, where the rescued pair came once
more into contact with civilization, Dobrinton was bitten by a dog which was
assumed to be mad, though it may only have been indiscriminating. (Saki)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
10:35 a.m.
Ar an tríú lá is fiche de mí Meitheamh, scríobh Stephen J. Turnbull:
Aidan, have any ideas offhand? Do you think stuffing ISO 8859/15 in
the search order immediately after ISO 8859/1 would be a good idea and
easy to do?
Here’s a patch to do it, but after ISO 8859/2, not /1, since we want ž and š
to be taken from the former.
Index: lisp/mule/general-late.el
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/mule/general-late.el,v
retrieving revision 1.2
diff -u -u -r1.2 general-late.el
--- lisp/mule/general-late.el 2006/12/29 18:09:46 1.2
+++ lisp/mule/general-late.el 2007/06/23 09:17:51
＠＠ -55,7 +55,21 ＠＠
;; At this point in the dump, all the charsets have been loaded. Now, load
;; their Unicode mappings.
(if load-unicode-tables-at-dump-time
- (let ((data-directory (expand-file-name "etc" source-directory)))
- (load-unicode-tables)))
+ (let ((data-directory (expand-file-name "etc" source-directory))
+ latin-9-moved latin-2-and-after)
+ (load-unicode-tables)
+
+ ;; Move latin-iso8859-15 after latin-iso8859-2, in order to prefer its
+ ;; version of the Euro sign over that in the Greek fonts.
+ (setq latin-9-moved
+ (delq 'latin-iso8859-15
+ (mapcar 'charset-name (unicode-precedence-list)))
+ latin-2-and-after (memq 'latin-iso8859-2 latin-9-moved))
+ (when (and latin-9-moved (find-charset 'latin-iso8859-15))
+ (setcdr latin-2-and-after (cons 'latin-iso8859-15
+ (cdr latin-2-and-after)))
+ (debug-print "latin-9-moved is %S, latin-2-and-after %S"
+ latin-9-moved latin-2-and-after)
+ (set-default-unicode-precedence-list latin-9-moved))))
;;; general-late.el ends here
Index: lisp/mule/mule-cmds.el
===================================================================
RCS file: /pack/xemacscvs/XEmacs/xemacs/lisp/mule/mule-cmds.el,v
retrieving revision 1.32
diff -u -u -r1.32 mule-cmds.el
--- lisp/mule/mule-cmds.el 2007/06/03 17:32:38 1.32
+++ lisp/mule/mule-cmds.el 2007/06/23 09:17:55
＠＠ -1486,7 +1486,20 ＠＠
"Initialize MULE environment at startup. Don't call this."
(when (not load-unicode-tables-at-dump-time)
- (load-unicode-tables))
+ (load-unicode-tables)
+
+ ;; Move latin-iso8859-15 after latin-iso8859-2, in order to prefer its
+ ;; version of the Euro sign to that for Greek.
+ (let* ((latin-9-moved
+ (delq 'latin-iso8859-15
+ (mapcar 'charset-name (unicode-precedence-list))))
+ (latin-2-and-after (memq 'latin-iso8859-2 latin-9-moved)))
+ (when (and latin-9-moved (find-charset 'latin-iso8859-15))
+ (setcdr latin-2-and-after (cons 'latin-iso8859-15
+ (cdr latin-2-and-after)))
+ ; (debug-print "latin-9-moved is %S, latin-2-and-after %S"
+ ; latin-9-moved latin-2-and-after)
+ (set-default-unicode-precedence-list latin-9-moved))))
;; This is called (currently; might be moved earlier) from startup.el,
;; after the basic GUI systems have been initialized, and just before the
--
On the quay of the little Black Sea port, where the rescued pair came once
more into contact with civilization, Dobrinton was bitten by a dog which was
assumed to be mad, though it may only have been indiscriminating. (Saki)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
6145
days inactive
6146
days old
6 comments
3 participants
participants (3)
-
Aidan Kehoe -
Mike FABIAN -
Stephen J. Turnbull