11:46 p.m.
It's a case-fold-search issue I haven't been able to resolve in
half-an-hour.
%S replacement is done, even though patcher-logmsg-commit-command
%contains %s.
I was able to commit my pending change using patcher.el<1.11>.
case treatment in
(defun replace-in-string (str regexp newtext &optional literal)
is on trial.
It uses re-search-forward for strings longer than 50, string-match otherwise.
case treatment is undocumented for re-search-forward, depending on
buffer argument for string-match.
(if (> (length str) 50)
(with-temp-buffer
(insert str)
(goto-char 1)
(while (re-search-forward regexp nil t)
(replace-match newtext t literal))
(buffer-string))
(let ((start 0) newstr)
(while (string-match regexp str start)
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
9:54 a.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
Hi again!
replace-in-string implements an arbitrary limit bug that would turn
some guys in Redwood green with envy :-)
The second r-i-s does NOT honor case-fold-search in
(emacs-version)
"XEmacs 21.4 (patch 9) \"Informed Management (RC1)\" [Lucid]
(i586-pc-win32) of Sat May 25 2002 on D5DC120J"
"XEmacs 21.5 (beta7) \"broccoflower\" [Lucid] (i586-pc-win32, Mule) of Sat
Jul 06 2002 on D5DC120J"
(with-temp-buffer
(let (case-fold-search)
(message
"%s"
(replace-in-string
"%s under 50 characters in length"
"%S"
"STRING"
t))
(sit-for 3)
(message
"%s"
(replace-in-string
"This %s, on the other hand, is well over 50 characters in length"
"%S"
"STRING"
t))))
It does, however work correctly in
"XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid] (i386-pc-win32) of Mon
May 21 2001 on D5DC120J"
Who has the inside story?
Best regards,
Adrian
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
APA> It's a case-fold-search issue I haven't been able to resolve in
APA> half-an-hour.
APA> %S replacement is done, even though patcher-logmsg-commit-command
APA> %contains %s.
APA> I was able to commit my pending change using patcher.el<1.11>.
APA> case treatment in
APA> (defun replace-in-string (str regexp newtext &optional literal)
APA> is on trial.
APA> It uses re-search-forward for strings longer than 50, string-match otherwise.
APA> case treatment is undocumented for re-search-forward, depending on
APA> buffer argument for string-match.
APA> (if (> (length str) 50)
APA> (with-temp-buffer
APA> (insert str)
APA> (goto-char 1)
APA> (while (re-search-forward regexp nil t)
APA> (replace-match newtext t literal))
APA> (buffer-string))
APA> (let ((start 0) newstr)
APA> (while (string-match regexp str start)
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
10:52 a.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
Adrian Aichner wrote:
replace-in-string implements an arbitrary limit bug that would turn
some
guys in Redwood green with envy :-)
The second r-i-s does NOT honor case-fold-search in (emacs-version) "XEmacs
21.4 (patch 9) \"Informed Management (RC1)\" [Lucid] (i586-pc-win32) of Sat
May 25 2002 on D5DC120J" "XEmacs 21.5 (beta7) \"broccoflower\"
[Lucid]
(i586-pc-win32, Mule) of Sat Jul 06 2002 on D5DC120J"
It does, however work correctly in "XEmacs 21.1 (patch 14)
\"Cuyahoga
Valley\" [Lucid] (i386-pc-win32) of Mon May 21 2001 on D5DC120J"
Who has the inside story?
I think it's related to this ChangeLog entry:
* subr.el (replace-in-string):
Rewrite this function to avoid N^2 behavior with large strings --
catastrophic with the new Windows selection code! (Apparently the
author of this function didn't realize there was a fun
replace-match that could make his life much easier, because we
duplicated the entire logic. The new version is smaller, easier
to understand, much more robust, and has extended features --
those of replace-match.)
Given the size of the entry, and the tone it's written in, one does not find
it difficult to figure out the author ..... :-)
2000-07-30 Ben Wing <ben(a)xemacs.org>
The problem we have is that re-search-forward does not currently honor
case-fold-search.
--
Didier Verna, didier(a)lrde.epita.fr, http://www.lrde.epita.fr/~didier
EPITA / LRDE, 14-16 rue Voltaire Tel.+33 (1) 53 14 59 47
94276 Le Kremlin-Bicêtre, France Fax.+33 (1) 53 14 59 22 didier(a)xemacs.org
12:12 p.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
>>>> "drv" == Didier Verna
<didier(a)xemacs.org> writes:
drv> Adrian Aichner wrote:
> replace-in-string implements an arbitrary limit bug that would
turn some
> guys in Redwood green with envy :-)
>
> The second r-i-s does NOT honor case-fold-search in (emacs-version) "XEmacs
> 21.4 (patch 9) \"Informed Management (RC1)\" [Lucid] (i586-pc-win32) of
Sat
> May 25 2002 on D5DC120J" "XEmacs 21.5 (beta7) \"broccoflower\"
[Lucid]
> (i586-pc-win32, Mule) of Sat Jul 06 2002 on D5DC120J"
> It does, however work correctly in "XEmacs 21.1 (patch 14)
\"Cuyahoga
> Valley\" [Lucid] (i386-pc-win32) of Mon May 21 2001 on D5DC120J"
>
> Who has the inside story?
drv> I think it's related to this ChangeLog entry:
drv> * subr.el (replace-in-string):
drv> Rewrite this function to avoid N^2 behavior with large strings --
drv> catastrophic with the new Windows selection code! (Apparently the
drv> author of this function didn't realize there was a fun
drv> replace-match that could make his life much easier, because we
drv> duplicated the entire logic. The new version is smaller, easier
drv> to understand, much more robust, and has extended features --
drv> those of replace-match.)
Seems like the re-writer did not realize a few things either :-)
drv> Given the size of the entry, and the tone it's written in,
drv> one does not find it difficult to figure out the author
drv> ..... :-)
drv> 2000-07-30 Ben Wing <ben(a)xemacs.org>
drv> The problem we have is that re-search-forward does not
drv> currently honor case-fold-search.
It does!
However, the use of (with-temp-buffer ...) for strings > 50 chars (42
I would understand, but 50, yuck!) in replace-in-string makes is use
the default-value of case-fold-search!
Didier, both of us seem to be fans of case-fold-search t, yes?
But hey, patcher will work for people who have it set to nil :-)
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
9:04 a.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
Adrian Aichner wrote:
drv> The problem we have is that re-search-forward does not
drv> currently honor case-fold-search.
It does!
However, the use of (with-temp-buffer ...) for strings > 50 chars (42
I would understand, but 50, yuck!) in replace-in-string makes is use
the default-value of case-fold-search!
You're right. Fix comin'up
Didier, both of us seem to be fans of case-fold-search t, yes?
Right.
--
Didier Verna, didier(a)lrde.epita.fr, http://www.lrde.epita.fr/~didier
EPITA / LRDE, 14-16 rue Voltaire Tel.+33 (1) 53 14 59 47
94276 Le Kremlin-Bicêtre, France Fax.+33 (1) 53 14 59 22 didier(a)xemacs.org
7:55 a.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
Hi Vin, Didier,
21.1.14 shows the buggy behavior for short and long strings!
(emacs-version)
"XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid] (i386-pc-win32) of Mon
May 21 2001 on D5DC120J"
entertains this bug for strings of any length:
(replace-in-string
"This %s has under 50 characters." "%S" "string")
"This string has under 50 characters."
(replace-in-string
"This %s, on the other hand, has just over 50 characters." "%S"
"string")
"This string, on the other hand, has just over 50 characters."
Didier, we might need to work around this bug in patcher.el.
What do you think?
Adrian
APA> Hi again!
APA> replace-in-string implements an arbitrary limit bug that would turn
APA> some guys in Redwood green with envy :-)
APA> The second r-i-s does NOT honor case-fold-search in
APA> (emacs-version)
APA> "XEmacs 21.4 (patch 9) \"Informed Management (RC1)\" [Lucid]
(i586-pc-win32) of Sat May 25 2002 on D5DC120J"
APA> "XEmacs 21.5 (beta7) \"broccoflower\" [Lucid] (i586-pc-win32,
Mule) of Sat Jul 06 2002 on D5DC120J"
APA> (with-temp-buffer
APA> (let (case-fold-search)
APA> (message
APA> "%s"
APA> (replace-in-string
APA> "%s under 50 characters in length"
APA> "%S"
APA> "STRING"
APA> t))
APA> (sit-for 3)
APA> (message
APA> "%s"
APA> (replace-in-string
APA> "This %s, on the other hand, is well over 50 characters in
length"
APA> "%S"
APA> "STRING"
APA> t))))
APA> It does, however work correctly in
APA> "XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid]
(i386-pc-win32) of Mon May 21 2001 on D5DC120J"
APA> Who has the inside story?
APA> Best regards,
APA> Adrian
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
APA> It's a case-fold-search issue I haven't been able to resolve in
APA> half-an-hour.
APA> %S replacement is done, even though patcher-logmsg-commit-command
APA> %contains %s.
APA> I was able to commit my pending change using patcher.el<1.11>.
APA> case treatment in
APA> (defun replace-in-string (str regexp newtext &optional literal)
APA> is on trial.
APA> It uses re-search-forward for strings longer than 50, string-match otherwise.
APA> case treatment is undocumented for re-search-forward, depending on
APA> buffer argument for string-match.
APA> (if (> (length str) 50)
APA> (with-temp-buffer
APA> (insert str)
APA> (goto-char 1)
APA> (while (re-search-forward regexp nil t)
APA> (replace-match newtext t literal))
APA> (buffer-string))
APA> (let ((start 0) newstr)
APA> (while (string-match regexp str start)
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
8:46 a.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
Adrian Aichner wrote:
21.1.14 shows the buggy behavior for short and long strings!
(emacs-version) "XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid]
(i386-pc-win32) of Mon May 21 2001 on D5DC120J" entertains this bug for
strings of any length:
(replace-in-string
"This %s has under 50 characters." "%S" "string")
"This string has under 50 characters."
(replace-in-string
"This %s, on the other hand, has just over 50 characters." "%S"
"string")
"This string, on the other hand, has just over 50 characters."
But what's the value of case-fold-search in that case ?
Didier, we might need to work around this bug in patcher.el.
What do you think?
I'm a bit reluctant to "fix" Patcher when the bug is in XEmacs. I
would definitely not for unstable versions like 21.4 and 21.5. The question
sounds reasonable for 21.1 which is the official stable version. But how many
users of 21.1 are likely to use Patcher right now ?
--
Didier Verna, didier(a)lrde.epita.fr, http://www.lrde.epita.fr/~didier
EPITA / LRDE, 14-16 rue Voltaire Tel.+33 (1) 53 14 59 47
94276 Le Kremlin-Bicêtre, France Fax.+33 (1) 53 14 59 22 didier(a)xemacs.org
3:38 p.m.
New subject: [WICKED BREAKAGE] replace-in-string (was: patcher.el<1.12, 1.13> have broken patcher-logmsg-commit)
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
Sorry for the false alarm!
(let
(case-fold-search)
(message "%s"
(replace-in-string
"This %s has under 50 characters." "%S"
"string"))
(sit-for 3)
(message "%s"
(replace-in-string
"This %s, on the other hand, has just over 50 characters."
"%S" "string")))
works fine!
Thanks for the hint, Didier!
Too-trigger-happily yours,
Adrian
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
APA> Hi Vin, Didier,
APA> 21.1.14 shows the buggy behavior for short and long strings!
APA> (emacs-version)
APA> "XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid]
(i386-pc-win32) of Mon May 21 2001 on D5DC120J"
APA> entertains this bug for strings of any length:
APA> (replace-in-string
APA> "This %s has under 50 characters." "%S"
"string")
APA> "This string has under 50 characters."
APA> (replace-in-string
APA> "This %s, on the other hand, has just over 50 characters."
"%S" "string")
APA> "This string, on the other hand, has just over 50 characters."
APA> Didier, we might need to work around this bug in patcher.el.
APA> What do you think?
APA> Adrian
APA> Hi again!
APA> replace-in-string implements an arbitrary limit bug that would turn
APA> some guys in Redwood green with envy :-)
APA> The second r-i-s does NOT honor case-fold-search in
APA> (emacs-version)
APA> "XEmacs 21.4 (patch 9) \"Informed Management (RC1)\" [Lucid]
(i586-pc-win32) of Sat May 25 2002 on D5DC120J"
APA> "XEmacs 21.5 (beta7) \"broccoflower\" [Lucid] (i586-pc-win32,
Mule) of Sat Jul 06 2002 on D5DC120J"
APA> (with-temp-buffer
APA> (let (case-fold-search)
APA> (message
APA> "%s"
APA> (replace-in-string
APA> "%s under 50 characters in length"
APA> "%S"
APA> "STRING"
APA> t))
APA> (sit-for 3)
APA> (message
APA> "%s"
APA> (replace-in-string
APA> "This %s, on the other hand, is well over 50 characters in length"
APA> "%S"
APA> "STRING"
APA> t))))
APA> It does, however work correctly in
APA> "XEmacs 21.1 (patch 14) \"Cuyahoga Valley\" [Lucid]
(i386-pc-win32) of Mon May 21 2001 on D5DC120J"
APA> Who has the inside story?
APA> Best regards,
APA> Adrian
>>>> "APA" == Adrian Aichner
<adrian(a)xemacs.org> writes:
APA> It's a case-fold-search issue I haven't been able to resolve in
APA> half-an-hour.
APA> %S replacement is done, even though patcher-logmsg-commit-command
APA> %contains %s.
APA> I was able to commit my pending change using patcher.el<1.11>.
APA> case treatment in
APA> (defun replace-in-string (str regexp newtext &optional literal)
APA> is on trial.
APA> It uses re-search-forward for strings longer than 50, string-match otherwise.
APA> case treatment is undocumented for re-search-forward, depending on
APA> buffer argument for string-match.
APA> (if (> (length str) 50)
APA> (with-temp-buffer
APA> (insert str)
APA> (goto-char 1)
APA> (while (re-search-forward regexp nil t)
APA> (replace-match newtext t literal))
APA> (buffer-string))
APA> (let ((start 0) newstr)
APA> (while (string-match regexp str start)
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
APA> --
APA> Adrian Aichner
APA> mailto:adrian＠xemacs.org
APA> http://www.xemacs.org/
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
10:22 a.m.
Adrian Aichner wrote:
It's a case-fold-search issue I haven't been able to resolve
in
half-an-hour.
%S replacement is done, even though patcher-logmsg-commit-command
%contains %s.
Even in 1.13 (patcher 3.3.1) ?? It was supposed to fix this bug.
(replace-in-string "foo%sbar" "%S" " ") => "foo
bar"
(let (case-fold-search)
(replace-in-string "foo%sbar" "%S" " ")) =>
"foo%sbar"
That's how I fixed it.
I was able to commit my pending change using patcher.el<1.11>.
case treatment in
(defun replace-in-string (str regexp newtext &optional literal)
is on trial.
It uses re-search-forward for strings longer than 50, string-match otherwise.
Hmmmm. I had not tested strings longer than 50 characters, but you
found a bug here:
(let (case-fold-search)
(replace-in-string "foo%sbar" "%S" " "))
=> "foo%sbar"
(let (case-fold-search)
(replace-in-string
"foo%sbar zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz" "%S"
" "))
=> "foo bar zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"
--
Didier Verna, didier(a)lrde.epita.fr, http://www.lrde.epita.fr/~didier
EPITA / LRDE, 14-16 rue Voltaire Tel.+33 (1) 53 14 59 47
94276 Le Kremlin-Bicêtre, France Fax.+33 (1) 53 14 59 22 didier(a)xemacs.org
7960
days inactive
7963
days old
8 comments
2 participants
participants (2)
-
Adrian Aichner -
Didier Verna