Below is the emacs-devel thread about adding an isearch keybinding to yank one letter at a time into the search string. If you're in a hurry, you can get a good summary just by reading the last two messages, from Karl Fogel and Gerd Moellman respectively.
From: Karl Fogel <kfogel@collab.net>
To: emacs-devel@gnu.org
Subject: [PATCH] better isearch support for complex input methods
Date: 02 Apr 2001 20:45:46 -0500
Reply-to: kfogel@red-bean.com
Message-ID: <87pueueoxh.fsf@collab.net>
This patch adds a new binding to isearch mode, C-o, which grabs one
letter at a time from the buffer and adds it to the search string.
Here's why this is useful:
Until recently, there was no point having a special keybinding for
this in isearch mode, because it would almost always have been just as
easy to type the letter itself, as to type a special binding to grab
the letter.
However, with modern input methods (such as `chinese-tonepy'), typing
a letter can be an involved process. In fact, in Chinese language
environments, what Emacs thinks of as a single letter is often
conceptually a whole word, i.e., one ideograph. It's very useful to
be able to grab one ideograph at a time from the buffer and add it to
the search string.
Currently, if one uses C-w in isearch in Chinese text, it grabs to the
next punctuation mark or end of line. This is usually not what the
user wanted -- it's much more likely that they want to grab one
Chinese character at a time.
This patch binds C-o to a new function `isearch-yank-letter'. C-o
seemed like a key unlikely to be often used for exiting isearch.
There may be a better binding I haven't thought of, though.
-Karl
2001-04-02 Karl Fogel <kfogel@red-bean.com>
* isearch.el (isearch-mode-map): Bind C-o in isearch to yank the
next letter from the buffer into the search string.
(isearch-yank-internal): New helper function, contains common
internals of next three.
(isearch-yank-letter): New function.
(isearch-yank-word): Implement using isearch-yank-internal.
(isearch-yank-line): Implement using isearch-yank-internal.
Index: isearch.el
===================================================================
RCS file: /cvs/emacs/lisp/isearch.el,v
retrieving revision 1.188
diff -u -r1.188 isearch.el
--- isearch.el 2001/02/05 17:13:28 1.188
+++ isearch.el 2001/04/03 00:20:57
@@ -286,6 +286,7 @@
(define-key map " " 'isearch-whitespace-chars)
(define-key map [?\S-\ ] 'isearch-whitespace-chars)
+ (define-key map "\C-o" 'isearch-yank-letter)
(define-key map "\C-w" 'isearch-yank-word)
(define-key map "\C-y" 'isearch-yank-line)
@@ -1057,24 +1058,32 @@
(isearch-yank-x-selection)
(mouse-yank-at-click click arg))))
-(defun isearch-yank-word ()
- "Pull next word from buffer into search string."
- (interactive)
+(defun isearch-yank-internal (jumpform)
+ "Pull the text from point to the point reached by JUMPFORM.
+JUMPFORM is a lambda expression that takes no arguments and returns a
+buffer position, possibly having moved point to that position. For
+example, it might move point forward by a word and return point, or it
+might return the position of the end of the line."
(isearch-yank-string
(save-excursion
(and (not isearch-forward) isearch-other-end
(goto-char isearch-other-end))
- (buffer-substring-no-properties
- (point) (progn (forward-word 1) (point))))))
+ (buffer-substring-no-properties (point) (funcall jumpform)))))
+(defun isearch-yank-letter ()
+ "Pull next letter from buffer into search string."
+ (interactive)
+ (isearch-yank-internal (lambda () (forward-char 1) (point))))
+
+(defun isearch-yank-word ()
+ "Pull next word from buffer into search string."
+ (interactive)
+ (isearch-yank-internal (lambda () (forward-word 1) (point))))
+
(defun isearch-yank-line ()
"Pull rest of line from buffer into search string."
(interactive)
- (isearch-yank-string
- (save-excursion
- (and (not isearch-forward) isearch-other-end
- (goto-char isearch-other-end))
- (buffer-substring-no-properties (point) (line-end-position)))))
+ (isearch-yank-internal (lambda () (line-end-position))))
(defun isearch-search-and-update ()
From: Kenichi Handa <handa@etl.go.jp>
To: kfogel@red-bean.com
CC: emacs-devel@gnu.org
Subject: Re: [PATCH] better isearch support for complex input methods
Message-Id: <200104030440.NAA02762@mule.m17n.org>
Karl Fogel <kfogel@collab.net> writes:
> This patch adds a new binding to isearch mode, C-o, which grabs one
> letter at a time from the buffer and adds it to the search string.
I surely like this feature. Actually, it's in my todo list.
> Here's why this is useful:
> Until recently, there was no point having a special keybinding for
> this in isearch mode, because it would almost always have been just as
> easy to type the letter itself, as to type a special binding to grab
> the letter.
> However, with modern input methods (such as `chinese-tonepy'), typing
> a letter can be an involved process. In fact, in Chinese language
> environments, what Emacs thinks of as a single letter is often
> conceptually a whole word, i.e., one ideograph. It's very useful to
> be able to grab one ideograph at a time from the buffer and add it to
> the search string.
> Currently, if one uses C-w in isearch in Chinese text, it grabs to the
> next punctuation mark or end of line. This is usually not what the
> user wanted -- it's much more likely that they want to grab one
> Chinese character at a time.
> This patch binds C-o to a new function `isearch-yank-letter'. C-o
> seemed like a key unlikely to be often used for exiting isearch.
> There may be a better binding I haven't thought of, though.
> -Karl
> 2001-04-02 Karl Fogel <kfogel@red-bean.com>
> * isearch.el (isearch-mode-map): Bind C-o in isearch to yank the
> next letter from the buffer into the search string.
> (isearch-yank-internal): New helper function, contains common
> internals of next three.
> (isearch-yank-letter): New function.
> (isearch-yank-word): Implement using isearch-yank-internal.
> (isearch-yank-line): Implement using isearch-yank-internal.
> Index: isearch.el
> ===================================================================
> RCS file: /cvs/emacs/lisp/isearch.el,v
> retrieving revision 1.188
> diff -u -r1.188 isearch.el
> --- isearch.el 2001/02/05 17:13:28 1.188
> +++ isearch.el 2001/04/03 00:20:57
> @@ -286,6 +286,7 @@
> (define-key map " " 'isearch-whitespace-chars)
> (define-key map [?\S-\ ] 'isearch-whitespace-chars)
> + (define-key map "\C-o" 'isearch-yank-letter)
> (define-key map "\C-w" 'isearch-yank-word)
> (define-key map "\C-y" 'isearch-yank-line)
> @@ -1057,24 +1058,32 @@
> (isearch-yank-x-selection)
> (mouse-yank-at-click click arg))))
> -(defun isearch-yank-word ()
> - "Pull next word from buffer into search string."
> - (interactive)
> +(defun isearch-yank-internal (jumpform)
> + "Pull the text from point to the point reached by JUMPFORM.
> +JUMPFORM is a lambda expression that takes no arguments and returns a
> +buffer position, possibly having moved point to that position. For
> +example, it might move point forward by a word and return point, or it
> +might return the position of the end of the line."
> (isearch-yank-string
> (save-excursion
> (and (not isearch-forward) isearch-other-end
> (goto-char isearch-other-end))
> - (buffer-substring-no-properties
> - (point) (progn (forward-word 1) (point))))))
> + (buffer-substring-no-properties (point) (funcall jumpform)))))
> +(defun isearch-yank-letter ()
> + "Pull next letter from buffer into search string."
> + (interactive)
> + (isearch-yank-internal (lambda () (forward-char 1) (point))))
> +
> +(defun isearch-yank-word ()
> + "Pull next word from buffer into search string."
> + (interactive)
> + (isearch-yank-internal (lambda () (forward-word 1) (point))))
> +
> (defun isearch-yank-line ()
> "Pull rest of line from buffer into search string."
> (interactive)
> - (isearch-yank-string
> - (save-excursion
> - (and (not isearch-forward) isearch-other-end
> - (goto-char isearch-other-end))
> - (buffer-substring-no-properties (point) (line-end-position)))))
> + (isearch-yank-internal (lambda () (line-end-position))))
> (defun isearch-search-and-update ()
> _______________________________________________
> Emacs-devel mailing list
> Emacs-devel@gnu.org
> http://mail.gnu.org/mailman/listinfo/emacs-devel
From: Kenichi Handa <handa@etl.go.jp> To: kfogel@red-bean.com CC: emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: Tue, 3 Apr 2001 13:45:23 +0900 (JST) Message-Id: <200104030445.NAA02860@mule.m17n.org> Karl Fogel <kfogel@collab.net> writes: > This patch adds a new binding to isearch mode, C-o, which grabs one > letter at a time from the buffer and adds it to the search string. I surely like this feature. Actually, I'm using the similar command personally, and committing it is in my todo list for 21.2. But, the command name should be isearch-yank-char (or isearch-yank-character) because it pulls not only letters but also newline, tab, space, etc. And, I think C-c is the better binding. --- Ken'ichi HANDA handa@etl.go.jp
From: Andreas Schwab <schwab@suse.de> To: Kenichi Handa <handa@etl.go.jp> Cc: kfogel@red-bean.com, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 03 Apr 2001 11:16:19 +0200 References: <200104030445.NAA02860@mule.m17n.org> Message-ID: <je3dbqqr6k.fsf@hawking.suse.de> Kenichi Handa <handa@etl.go.jp> writes: |> And, I think C-c is the better binding. I don't think so. C-c has already a defined meaning as a prefix key and should not be overloaded by isearch. Andreas. -- Andreas Schwab "And now for something SuSE Labs completely different." Andreas.Schwab@suse.de SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
From: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> To: Karl Fogel <kfogel@collab.net> Cc: emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: Tue, 03 Apr 2001 10:11:27 -0400 References: <87pueueoxh.fsf@collab.net> Message-Id: <200104031411.f33EBRO32119@rum.cs.yale.edu> > Currently, if one uses C-w in isearch in Chinese text, it grabs to the > next punctuation mark or end of line. This is usually not what the > user wanted -- it's much more likely that they want to grab one > Chinese character at a time. This suggests that instead of introducing a new binding C-o we should just change C-w so that it yanks in either one word or one char depending on the alphabet. Stefan
From: Karl Fogel <kfogel@collab.net> To: Andreas Schwab <schwab@suse.de> Cc: Kenichi Handa <handa@etl.go.jp>, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 03 Apr 2001 13:04:46 -0500 Message-ID: <874rw5hnb5.fsf@collab.net> References: <200104030445.NAA02860@mule.m17n.org> <je3dbqqr6k.fsf@hawking.suse.de> Reply-to: kfogel@red-bean.com Andreas Schwab <schwab@suse.de> writes: > |> And, I think C-c is the better binding. > > I don't think so. C-c has already a defined meaning as a prefix key and > should not be overloaded by isearch. I agree with Andreas. The important question is, which binding is _least_ likely to be used by people as an "exit isearch and start doing something else" key. C-o meets this test better than C-c, because C-c is the start of many commands, whereas C-o is the start of just one command, `open-line'. open-line is a very unlikely command to exit isearch mode with, because you usually don't open a new line immediately on having reached a search match. After all, most of the time, you're at some random point in the match, depending on how many characters you needed to type to reach the desired match. Therefore, the chances are low that it would also be an appropriate spot to open a new line. Kenichi Handa <handa@etl.go.jp> writes: > 21.2. But, the command name should be isearch-yank-char (or > isearch-yank-character) because it pulls not only letters > but also newline, tab, space, etc. Good point, thanks. I will change it to `isearch-yank-char'. Does anyone have a problem with my committing this change, then? -Karl
From: Karl Fogel <kfogel@collab.net> To: "Stefan Monnier" <monnier+gnu/emacs@rum.cs.yale.edu> Cc: emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 03 Apr 2001 13:48:34 -0500 Message-ID: <87elv9g6pp.fsf@collab.net> References: <87pueueoxh.fsf@collab.net> <200104031411.f33EBRO32119@rum.cs.yale.edu> Reply-to: kfogel@red-bean.com "Stefan Monnier" <monnier+gnu/emacs@rum.cs.yale.edu> writes: > > Currently, if one uses C-w in isearch in Chinese text, it grabs to the > > next punctuation mark or end of line. This is usually not what the > > user wanted -- it's much more likely that they want to grab one > > Chinese character at a time. > > This suggests that instead of introducing a new binding C-o we should > just change C-w so that it yanks in either one word or one char > depending on the alphabet. I would prefer that too, but I'm not sure how to implement it. Remember, the buffer may contain mixed-language text. We still want C-w to work right for multicharacter words. For example, imagine this text, where "X", "Y", and "Z" are Chinese characters: X Y is my name, and Z is my game If point is on X, someone might want to search the next occurrence of "X Y is my name". *Ideally*, they would type C-s C-w C-w C-w C-w C-w C-s However, this can only work if C-w knows to behave differently when looking at a multibyte Chinese character than when looking at some other kind of text. And of course, it can't just be for Chinese: Emacs would have to recognize a general classification of "characters that are themselves words" in various lanuages. Can it do this? I agree it would be a wonderfully usable solution, better than the C-o solution. I'm not sure how to implement it, though. Any ideas? [In the meantime, C-o is a vast improvement over what we have now, and it is of general use, even in non-multibyte-char buffers -- I've wanted to pull one char at a time into my search string before, without having to find the chars on my keyboard.] -Karl
From: "Eli Zaretskii" <eliz@is.elta.co.il> To: kfogel@red-bean.com CC: monnier+gnu/emacs@rum.cs.yale.edu, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Reply-to: Eli Zaretskii <eliz@is.elta.co.il> Message-Id: <8011-Tue03Apr2001205856+0200-eliz@is.elta.co.il> References: <87pueueoxh.fsf@collab.net> <200104031411.f33EBRO32119@rum.cs.yale.edu> <87elv9g6pp.fsf@collab.net> > From: Karl Fogel <kfogel@collab.net> > Date: 03 Apr 2001 13:48:34 -0500 > > However, this can only work if C-w knows to behave differently when > looking at a multibyte Chinese character than when looking at some > other kind of text. And of course, it can't just be for Chinese: > Emacs would have to recognize a general classification of "characters > that are themselves words" in various lanuages. Can it do this? Well, how about if it always treated entire charsets in the same manner? For example, all Chinese characters would be treated as words. We can surely classify characters by the character set they belong to, even when the internal character representation is changed to be based on Unicode.
From: "Eli Zaretskii" <eliz@is.elta.co.il> Date: Tue, 03 Apr 2001 21:03:55 +0200 To: kfogel@red-bean.com CC: emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Reply-to: Eli Zaretskii <eliz@is.elta.co.il> Message-Id: <1438-Tue03Apr2001210354+0200-eliz@is.elta.co.il> References: <200104030445.NAA02860@mule.m17n.org> <je3dbqqr6k.fsf@hawking.suse.de> <874rw5hnb5.fsf@collab.net> > From: Karl Fogel <kfogel@collab.net> > Date: 03 Apr 2001 13:04:46 -0500 > > Does anyone have a problem with my committing this change, then? If you intend to commit this now (i.e. before v21.1 is released), I think you need Gerd's approval, since we are under a feature freeze for quite some time.
From: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu>
To: "Eli Zaretskii" <eliz@is.elta.co.il>
Cc: kfogel@red-bean.com, emacs-devel@gnu.org
Subject: Re: [PATCH] better isearch support for complex input methods
Date: Tue, 03 Apr 2001 15:39:57 -0400
Message-Id: <200104031939.f33Jdvg01141@rum.cs.yale.edu>
References: <87pueueoxh.fsf@collab.net>
<200104031411.f33EBRO32119@rum.cs.yale.edu>
<87elv9g6pp.fsf@collab.net>
<8011-Tue03Apr2001205856+0200-eliz@is.elta.co.il>
> > From: Karl Fogel <kfogel@collab.net>
> > Date: 03 Apr 2001 13:48:34 -0500
> >
> > However, this can only work if C-w knows to behave differently when
> > looking at a multibyte Chinese character than when looking at some
> > other kind of text. And of course, it can't just be for Chinese:
> > Emacs would have to recognize a general classification of "characters
> > that are themselves words" in various lanuages. Can it do this?
>
> Well, how about if it always treated entire charsets in the same
> manner? For example, all Chinese characters would be treated as
> words. We can surely classify characters by the character set they
> belong to, even when the internal character representation is changed
> to be based on Unicode.
That's what I was thinking. It's probably not 100% correct, but
should be fairly close. Also it should be fairly easy to list
the relevant charsets since I would expect that all the ideogram-based
charsets are 2-byte charsets and we don't have thousands of those, do we ?
Stefan
From: Miles Bader <miles@lsi.nec.co.jp> To: Kenichi Handa <handa@etl.go.jp> Cc: kfogel@red-bean.com, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 03 Apr 2001 14:01:47 +0900 Reply-To: Miles Bader <miles@gnu.org> Message-ID: <buoy9ti8tl0.fsf@mcspd15.ucom.lsi.nec.co.jp> References: <200104030445.NAA02860@mule.m17n.org> Kenichi Handa <handa@etl.go.jp> writes: > And, I think C-c is the better binding. However, it's also normally a prefix command, so using it will block a much larger set of bindings from being used to exit isearch. C-o is currently used by quail's japanese input method to extend the translation region by one character, so there is _some_ precedent for using C-o. -Miles -- Love is a snowmobile racing across the tundra. Suddenly it flips over, pinning you underneath. At night the ice weasels come. --Nietzsche
From: Miles Bader <miles@lsi.nec.co.jp> To: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> Cc: Karl Fogel <kfogel@collab.net>, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 04 Apr 2001 10:02:09 +0900 Reply-To: Miles Bader <miles@gnu.org> Message-ID: <buohf05eaum.fsf@mcspd15.ucom.lsi.nec.co.jp> References: <87pueueoxh.fsf@collab.net> <200104031411.f33EBRO32119@rum.cs.yale.edu> "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> writes: > This suggests that instead of introducing a new binding C-o we should > just change C-w so that it yanks in either one word or one char > depending on the alphabet. Er, I'd like to use this functionality with ASCII too... (often C-w goes too far, and it'd be nice to just hit C-o 10 times instead of carefully typing in whatever happens to be at the match point) -Miles -- Love is a snowmobile racing across the tundra. Suddenly it flips over, pinning you underneath. At night the ice weasels come. --Nietzsche
From: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu>
To: Miles Bader <miles@lsi.nec.co.jp>
Cc: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu>,
Karl Fogel <kfogel@collab.net>, emacs-devel@gnu.org
Subject: Re: [PATCH] better isearch support for complex input methods
Date: Tue, 03 Apr 2001 22:43:26 -0400
References: <87pueueoxh.fsf@collab.net>
<200104031411.f33EBRO32119@rum.cs.yale.edu>
<buohf05eaum.fsf@mcspd15.ucom.lsi.nec.co.jp>
Message-Id: <200104040243.f342hQq02180@rum.cs.yale.edu>
> "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> writes:
> > This suggests that instead of introducing a new binding C-o we should
> > just change C-w so that it yanks in either one word or one char
> > depending on the alphabet.
>
> Er, I'd like to use this functionality with ASCII too... (often C-w goes
> too far, and it'd be nice to just hit C-o 10 times instead of carefully
> typing in whatever happens to be at the match point)
I use C-w M-e ... instead ;-)
In any case, I just realized that my earlier point about C-w moving
by either one word or one ideogram might actually apply not just
to C-w in isearch but to forward-word in general.
I'm not directly interested in such a change since I only ever use latin
charsets, but I'm still wondering...
Stefan
From: Miles Bader <miles@lsi.nec.co.jp> To: "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> Cc: Karl Fogel <kfogel@collab.net>, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 04 Apr 2001 11:55:05 +0900 Reply-To: Miles Bader <miles@gnu.org> Message-ID: <buod7ate5me.fsf@mcspd15.ucom.lsi.nec.co.jp> References: <87pueueoxh.fsf@collab.net> <200104031411.f33EBRO32119@rum.cs.yale.edu> <buohf05eaum.fsf@mcspd15.ucom.lsi.nec.co.jp> <200104040243.f342hQq02180@rum.cs.yale.edu> "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> writes: > In any case, I just realized that my earlier point about C-w moving > by either one word or one ideogram might actually apply not just > to C-w in isearch but to forward-word in general. Er, we already have C-f for that ... In the case of Japanese, at least, forward-word is still useful because it stops at kana/kanji boundaries. -Miles -- Love is a snowmobile racing across the tundra. Suddenly it flips over, pinning you underneath. At night the ice weasels come. --Nietzsche
From: Eli Zaretskii <eliz@is.elta.co.il>
To: Miles Bader <miles@lsi.nec.co.jp>
cc: Stefan Monnier <monnier+gnu/emacs@RUM.cs.yale.edu>,
Karl Fogel <kfogel@collab.net>, emacs-devel@gnu.org
Subject: Re: [PATCH] better isearch support for complex input methods
Date: Wed, 4 Apr 2001 09:40:24 +0200 (IST)
Message-ID: <Pine.SUN.3.91.1010404094000.23448G-100000@is>
On 4 Apr 2001, Miles Bader wrote:
> "Stefan Monnier" <monnier+gnu/emacs@flint.cs.yale.edu> writes:
> > This suggests that instead of introducing a new binding C-o we should
> > just change C-w so that it yanks in either one word or one char
> > depending on the alphabet.
>
> Er, I'd like to use this functionality with ASCII too...
We could have a user option for that.
From: Karl Fogel <kfogel@collab.net> To: "Stefan Monnier" <monnier+gnu/emacs@rum.cs.yale.edu> Cc: "Eli Zaretskii" <eliz@is.elta.co.il>, emacs-devel@gnu.org Subject: Re: [PATCH] better isearch support for complex input methods Date: 04 Apr 2001 11:41:41 -0500 Reply-to: kfogel@red-bean.com Message-ID: <87y9tgsjlm.fsf@collab.net> References: <87pueueoxh.fsf@collab.net> <200104031411.f33EBRO32119@rum.cs.yale.edu> <87elv9g6pp.fsf@collab.net> <8011-Tue03Apr2001205856+0200-eliz@is.elta.co.il> <200104031939.f33Jdvg01141@rum.cs.yale.edu> "Stefan Monnier" <monnier+gnu/emacs@rum.cs.yale.edu> writes: > > Well, how about if it always treated entire charsets in the same > > manner? For example, all Chinese characters would be treated as > > words. We can surely classify characters by the character set they > > belong to, even when the internal character representation is changed > > to be based on Unicode. > > That's what I was thinking. It's probably not 100% correct, but > should be fairly close. Also it should be fairly easy to list > the relevant charsets since I would expect that all the ideogram-based > charsets are 2-byte charsets and we don't have thousands of those, do we ? This sounds like a better plan than my original patch. I'm reading up on some multibyte stuff, and will post a new patch soon. Ideally Emacs would be able to "get it right" 100% of the time, instead of having a mostly-correct heuristic; will see if this is possible. Definitely won't commit it without asking first, thanks for reminding me about the feature freeze. It's no big deal if it doesn't make it into this release, IMHO. [Note: I abandoned this grander plan for lack of time, though it certainly would be a good thing to implement. See next message for why the existing patch is at least a lossless workaround. -Karl] -K