A few bugs not in the bug tracker (I think)

A few bugs not in the bug tracker (I think)

[Juanma Barranquero]

The following open bugs are not in the bug tracker, I think.

- face-valid-attribute-values chokes on font-family-list and new
font-*-table vars
  http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg00948.html

- ^M in the info files [Windows]
  http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg00739.html
  http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00215.html

- Disappearing ^J in ChangeLogs [Windows]
  http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg02050.html
  http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00075.html

- Crash displaying byte-code [Windows?]
  http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00236.html

BTW, how do you know whether a bug has already been reported, other
than perusing the open bug list? I see ways to search by bug #,
package, maintainer e-mail, severity, tag, etc., but not a text search
in the body of the messages (or even the subject).

 Juanma

Re: A few bugs not in the bug tracker (I think)

[Kenichi Handa]

In article <f7ccd24b0808060409x7317e093i6918324c0c057687@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> The following open bugs are not in the bug tracker, I think.
> - face-valid-attribute-values chokes on font-family-list and new
> font-*-table vars
>   http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg00948.html

I've just installed the fix.

> - ^M in the info files [Windows]
>   http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg00739.html
>   http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00215.html

The problem is the newly introduced null-byte detection
mechanism, but the discussion divergented to encoding of
non-ascii characters, and for the original problem of
null-byte in info file, no one replied to the question in
this mail:
  http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00225.html

> - Disappearing ^J in ChangeLogs [Windows]
>   http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg02050.html
>   http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00075.html

Could someone please find a reproducible recipe?

> - Crash displaying byte-code [Windows?]
>   http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00236.html

I have not yet investigated that problem.  At least, I can't
reproduce it on X.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: A few bugs not in the bug tracker (I think)

[Juanma Barranquero]

On Wed, Aug 6, 2008 at 14:31, Kenichi Handa <handa@m17n.org> wrote:

>> The following open bugs are not in the bug tracker, I think.
>> - face-valid-attribute-values chokes on font-family-list and new
>> font-*-table vars
>
> I've just installed the fix.

Thanks.

One thing which was not in the bug report is that, on Windows,

  (face-valid-attribute-values :stipple)

returns a list of all files in the current directory, because
x-bitmap-file-path is "." on Windows. That seems a bit absurd.

> The problem is the newly introduced null-byte detection
> mechanism, but the discussion divergented to encoding of
> non-ascii characters, and for the original problem of
> null-byte in info file, no one replied to the question in
> this mail:
>  http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00225.html

How would inhibit-null-byte-detection work for info files?

>> - Disappearing ^J in ChangeLogs [Windows]
>
> Could someone please find a reproducible recipe?

Wish I could. I've set up modification hooks in my Emacs to try and
find what is happening, or at least *when* it is happening.

>> - Crash displaying byte-code [Windows?]
>>   http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00236.html
>
> I have not yet investigated that problem.  At least, I can't
> reproduce it on X.

It might be related to font issues on Windows.

 Juanma

Re: A few bugs not in the bug tracker (I think)

[Don Armstrong]

On Wed, 06 Aug 2008, Juanma Barranquero wrote:
> BTW, how do you know whether a bug has already been reported, other
> than perusing the open bug list? I see ways to search by bug #,
> package, maintainer e-mail, severity, tag, etc., but not a text
> search in the body of the messages (or even the subject).

http://www.google.com/search?q=site%3Aemacsbugs.donarmstrong.com%20bytecode
 
for example, will search. There is also a full-text search engine, but
I haven't deployed it yet for emacs bugs. [I want to move things over
before I do that, but I'm currently at Debconf working on debbugs
enhancements, and haven't had a chance to do that.]


Don Armstrong

-- 
"The trouble with you, Ibid" he said, "is that you think you're the
biggest bloody authority on everything"
 -- Terry Pratchet _Pyramids_ p146

http://www.donarmstrong.com              http://rzlab.ucr.edu

Re: A few bugs not in the bug tracker (I think)

[Juanma Barranquero]

>> - face-valid-attribute-values chokes on font-family-list and new
>> font-*-table vars

BTW, speaking of face attributes... Is this (from bug #648) a bug?

Emacs 22.X:  (face-attribute 'default :inherit) => unspecified
Emacs 23.X:  (face-attribute 'default :inherit) => nil

  Juanma

Re: A few bugs not in the bug tracker (I think)

[Juanma Barranquero]

On Wed, Aug 6, 2008 at 15:35, Don Armstrong <don@donarmstrong.com> wrote:

> http://www.google.com/search?q=site%3Aemacsbugs.donarmstrong.com%20bytecode
>
> for example, will search.

Ah, I hadn't thought of that.

> There is also a full-text search engine, but
> I haven't deployed it yet for emacs bugs.

OK.

Thanks,

   Juanma

Re: A few bugs not in the bug tracker (I think)

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> Date: Wed, 06 Aug 2008 21:31:51 +0900
> Cc: emacs-devel@gnu.org
> 
> > - ^M in the info files [Windows]
> >   http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg00739.html
> >   http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00215.html
> 
> The problem is the newly introduced null-byte detection
> mechanism, but the discussion divergented to encoding of
> non-ascii characters, and for the original problem of
> null-byte in info file, no one replied to the question in
> this mail:
>   http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00225.html

I think the best solution would be that Emacs never tries to detect
whether Info files are binary.

Re: ^M in the info files

[Kenichi Handa]

I changed the Subject:

In article <f7ccd24b0808060614h4917f240oc261b8fedff0e8d0@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> > The problem is the newly introduced null-byte detection
> > mechanism, but the discussion divergented to encoding of
> > non-ascii characters, and for the original problem of
> > null-byte in info file, no one replied to the question in
> > this mail:
> >  http://lists.gnu.org/archive/html/emacs-devel/2008-07/msg00225.html

> How would inhibit-null-byte-detection work for info files?

For instance, by modifying info-insert-file-contents to bind
that variable to t while reading a file.

Eli Zaretskii <eliz@gnu.org> writes:

> I think the best solution would be that Emacs never tries to detect
> whether Info files are binary.

Do you have any idea to achieve that?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: ^M in the info files

[Eli Zaretskii]

> From: Kenichi Handa <handa@m17n.org>
> Date: Thu, 07 Aug 2008 09:48:37 +0900
> Cc: emacs-devel@gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I think the best solution would be that Emacs never tries to detect
> > whether Info files are binary.
> 
> Do you have any idea to achieve that?

Sorry, not yet.

Re: ^M in the info files

[Juanma Barranquero]

On Thu, Aug 7, 2008 at 02:48, Kenichi Handa <handa@m17n.org> wrote:

> For instance, by modifying info-insert-file-contents to bind
> that variable to t while reading a file.

Aha.

Well, that's a possible solution then. Is there any downside to it?

  Juanma

Re: ^M in the info files

[Kenichi Handa]

In article <f7ccd24b0808062022t71035360h5f120bcade3363e8@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> On Thu, Aug 7, 2008 at 02:48, Kenichi Handa <handa@m17n.org> wrote:
> > For instance, by modifying info-insert-file-contents to bind
> > that variable to t while reading a file.

> Aha.

> Well, that's a possible solution then. Is there any downside to it?

I don't see any bad effect to it.  But, as I'm not familiar
with the code of info, I don't know which place is the best
to bind that variable.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: A few bugs not in the bug tracker (I think)

[Miles Bader]

"Juanma Barranquero" <lekktu@gmail.com> writes:
> BTW, speaking of face attributes... Is this (from bug #648) a bug?
>
> Emacs 22.X:  (face-attribute 'default :inherit) => unspecified
> Emacs 23.X:  (face-attribute 'default :inherit) => nil

No.

The default face shouldn't have any unspecified attributes.

Well, I suppose you could consider it a bug in 22.x, though probably not
a very important one...

-Miles

-- 
Quack, n. A murderer without a license.

Re: A few bugs not in the bug tracker (I think)

[Juanma Barranquero]

On Thu, Aug 7, 2008 at 06:25, Miles Bader <miles.bader@necel.com> wrote:

> No.
>
> The default face shouldn't have any unspecified attributes.

OK.

> Well, I suppose you could consider it a bug in 22.x, though probably not
> a very important one...

Agreed.

  Juanma

Re: ^M in the info files

[Juanma Barranquero]

On Thu, Aug 7, 2008 at 05:47, Kenichi Handa <handa@m17n.org> wrote:

> I don't see any bad effect to it.  But, as I'm not familiar
> with the code of info, I don't know which place is the best
> to bind that variable.

I've tried implementing your suggestion; see the attached patch.

It works for uncompressed info files, and for compressed ones when
auto-compression-mode is off.
However, when auto-compression-mode is on, the conversion depends on
the settings of `file-coding-system-alist' for the given compression
method.

   Juanma



Index: src/coding.c
===================================================================
RCS file: /sources/emacs/emacs/src/coding.c,v
retrieving revision 1.390
diff -u -2 -r1.390 coding.c
--- src/coding.c	9 Jul 2008 13:05:56 -0000	1.390
+++ src/coding.c	7 Aug 2008 10:42:46 -0000
@@ -381,4 +381,7 @@
 int inhibit_iso_escape_detection;

+/* Flag to inhibit detection of binary files through null bytes.  */
+int inhibit_null_byte_detection;
+
 /* Flag to make buffer-file-coding-system inherit from process-coding.  */
 int inherit_process_coding_system;
@@ -5895,5 +5898,5 @@
 	  if (i < coding_category_raw_text)
 	    setup_coding_system (CODING_ID_NAME (this->id), coding);
-	  else if (null_byte_found)
+	  else if (null_byte_found && ! inhibit_null_byte_detection)
 	    setup_coding_system (Qno_conversion, coding);
 	  else if ((detect_info.rejected & CATEGORY_MASK_ANY)
@@ -10233,4 +10236,23 @@
   inhibit_iso_escape_detection = 0;

+  DEFVAR_BOOL ("inhibit-null-byte-detection",
+	       &inhibit_null_byte_detection,
+	       doc: /*
+If non-nil, Emacs ignores null bytes on code detection.
+
+By default, on reading a file, Emacs tries to detect how the text is
+encoded.  This code detection is sensitive to null bytes.  If the
+text contains null bytes, the file is determined as containing
+binary data.
+
+However, there may be a case that you want to read non-binary data
+that contains null bytes.  In such a case, you can set this variable
+to non-nil.
+
+The default value is nil, and it is strongly recommended not to change
+it.  This variable is intended to be bound to t in the few instances
+where that is useful, for example to read certain info files.  */);
+  inhibit_null_byte_detection = 0;
+
   DEFVAR_LISP ("translation-table-for-input", &Vtranslation_table_for_input,
 	       doc: /* Char table for translating self-inserting characters.
Index: lisp/info.el
===================================================================
RCS file: /sources/emacs/emacs/lisp/info.el,v
retrieving revision 1.540
diff -u -2 -r1.540 info.el
--- lisp/info.el	30 Jul 2008 17:16:46 -0000	1.540
+++ lisp/info.el	7 Aug 2008 11:20:36 -0000
@@ -456,5 +456,6 @@
 	    (apply 'call-process-region (point-min) (point-max)
 		   (car decoder) t t nil (cdr decoder))))
-      (insert-file-contents fullname visit))))
+      (let ((inhibit-null-byte-detection t))
+	(insert-file-contents fullname visit)))))
 \f
 (defun Info-default-dirs ()

Re: ^M in the info files

[Stefan Monnier]

>> How would inhibit-null-byte-detection work for info files?
> For instance, by modifying info-insert-file-contents to bind
> that variable to t while reading a file.

An alternative to inhibit-null-byte-detection would be something like
inhibit-coding-systems which would contain a list of coding systems (or
maybe coding categories?) we know we don't want to use.


        Stefan

Re: ^M in the info files

[Kenichi Handa]

In article <f7ccd24b0808070601i2897b57ak98d6343b2442f40c@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> On Thu, Aug 7, 2008 at 05:47, Kenichi Handa <handa@m17n.org> wrote:
> > I don't see any bad effect to it.  But, as I'm not familiar
> > with the code of info, I don't know which place is the best
> > to bind that variable.

> I've tried implementing your suggestion; see the attached patch.

Thank you!

> It works for uncompressed info files, and for compressed ones when
> auto-compression-mode is off.
> However, when auto-compression-mode is on, the conversion depends on
> the settings of `file-coding-system-alist' for the given compression
> method.

You handled inhibit_iso_escape_detection only in
detect_coding.  Isn't the problem fixed by modifing the
detect_coding_system too.  That funciton does code detection
when called from detect-coding-region/string.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: ^M in the info files

[Juanma Barranquero]

On Fri, Aug 8, 2008 at 03:28, Kenichi Handa <handa@m17n.org> wrote:

> You handled inhibit_iso_escape_detection only in
> detect_coding.  Isn't the problem fixed by modifing the
> detect_coding_system too.

Well, you give me too much credit; I just did the easiest change that
could possibly work :-) [It didn't]

But AFAICS, the trouble with compressed info files is not because of
null-byte detection. If I gzip efaq and then visit it through Info,
the buffer is detected as utf-8-unix, not no-conversion.

  Juanma

Re: ^M in the info files

[Kenichi Handa]

In article <f7ccd24b0808080402m2d2b958aq9fe7f82a25d7b48d@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> On Fri, Aug 8, 2008 at 03:28, Kenichi Handa <handa@m17n.org> wrote:
> > You handled inhibit_iso_escape_detection only in
> > detect_coding.  Isn't the problem fixed by modifing the
> > detect_coding_system too.

> Well, you give me too much credit; I just did the easiest change that
> could possibly work :-) [It didn't]

> But AFAICS, the trouble with compressed info files is not because of
> null-byte detection. If I gzip efaq and then visit it through Info,
> the buffer is detected as utf-8-unix, not no-conversion.

Do you mean that utf-8-unix is an incorrect detection?
Or, are there any other problems?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: ^M in the info files

[Juanma Barranquero]

On Wed, Aug 13, 2008 at 12:08, Kenichi Handa <handa@m17n.org> wrote:

> Do you mean that utf-8-unix is an incorrect detection?

Well, I would have expected that, if it detected the file as utf-8-dos
the CR/LF pairs would be correctly shown... (I'm talking of the case
of a gzipped Windows-generated info file).

   Juanma

RE: ^M in the info files

[Drew Adams]

> From: Stefan Monnier Sent: Thursday, August 07, 2008 1:19 PM
> >> How would inhibit-null-byte-detection work for info files?
> >
> > For instance, by modifying info-insert-file-contents to bind
> > that variable to t while reading a file.
> 
> An alternative to inhibit-null-byte-detection would be something like
> inhibit-coding-systems which would contain a list of coding 
> systems (or maybe coding categories?) we know we don't want to use.

Whatever happened to this thread and the associated bugs: #876, #1117, #1284?

It sounds like there were alternative proposals about how to fix this, but there
was no discussion to try to reach a consensus or a decision. Is that where
things were left?

Meanwhile, it's still impossible to use the index in Info manuals on Windows,
and it's impossible to use some manuals (e.g. Viper) at all.

Re: ^M in the info files

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Date: Fri, 28 Nov 2008 13:28:34 -0800
> Cc: 'Juanma Barranquero' <lekktu@gmail.com>, emacs-devel@gnu.org
> 
> Whatever happened to this thread and the associated bugs: #876, #1117, #1284?
> 
> It sounds like there were alternative proposals about how to fix this, but there
> was no discussion to try to reach a consensus or a decision. Is that where
> things were left?
> 
> Meanwhile, it's still impossible to use the index in Info manuals on Windows,
> and it's impossible to use some manuals (e.g. Viper) at all.

Don't worry, this will get fixed before Emacs 23 is ready for release.

I will work on it soon if no one beats me to it.

Re: ^M in the info files

[Juanma Barranquero]

On Fri, Nov 28, 2008 at 23:39, Eli Zaretskii <eliz@gnu.org> wrote:

> I will work on it soon if no one beats me to it.

How are you going to fix it?

  Juanma

RE: ^M in the info files

[Drew Adams]

> > Whatever happened to this thread and the associated bugs: 
> > #876, #1117, #1284?
> > 
> > It sounds like there were alternative proposals about how 
> > to fix this, but there was no discussion to try to reach a
> > consensus or a decision. Is that where things were left?
> > 
> > Meanwhile, it's still impossible to use the index in Info 
> > manuals on Windows, and it's impossible to use some manuals
> > (e.g. Viper) at all.
> 
> Don't worry, this will get fixed before Emacs 23 is ready for release.
> I will work on it soon if no one beats me to it.

OK, thanks.

But what will you do?
It sounds like nothing was decided wrt the fix that's needed.

Re: ^M in the info files

[Eli Zaretskii]

> Date: Fri, 28 Nov 2008 23:44:49 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> On Fri, Nov 28, 2008 at 23:39, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > I will work on it soon if no one beats me to it.
> 
> How are you going to fix it?

I don't know yet; figuring that out is part of the job.

Last time when I had a few hours to work on this, my Internet
connection died in the middle of reading past messages about this
problem, and came back up when the time I had was up.

Re: ^M in the info files

[Eli Zaretskii]

> From: "Drew Adams" <drew.adams@oracle.com>
> Cc: <monnier@iro.umontreal.ca>, <handa@m17n.org>, <lekktu@gmail.com>,
>         <emacs-devel@gnu.org>
> Date: Fri, 28 Nov 2008 14:49:32 -0800
> 
> > Don't worry, this will get fixed before Emacs 23 is ready for release.
> > I will work on it soon if no one beats me to it.
> 
> OK, thanks.
> 
> But what will you do?
> It sounds like nothing was decided wrt the fix that's needed.

Obviously, I will need first to decide which of the two suggested
approaches I like best.  Or maybe I will come up with some third
solution.  I will know when I actually start working on this.

Re: ^M in the info files

[Juanma Barranquero]

On Sat, Nov 29, 2008 at 11:50, Eli Zaretskii <eliz@gnu.org> wrote:

> Last time when I had a few hours to work on this, my Internet
> connection died in the middle of reading past messages about this
> problem, and came back up when the time I had was up.

An omen if I ever saw one...

  Juanma

Re: ^M in the info files

[Eli Zaretskii]

> Date: Sat, 29 Nov 2008 00:39:27 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: lekktu@gmail.com, emacs-devel@gnu.org, monnier@iro.umontreal.ca,
> 	handa@m17n.org
> 
> > From: "Drew Adams" <drew.adams@oracle.com>
> > Date: Fri, 28 Nov 2008 13:28:34 -0800
> > Cc: 'Juanma Barranquero' <lekktu@gmail.com>, emacs-devel@gnu.org
> > 
> > Whatever happened to this thread and the associated bugs: #876, #1117, #1284?
> > 
> > It sounds like there were alternative proposals about how to fix this, but there
> > was no discussion to try to reach a consensus or a decision. Is that where
> > things were left?
> > 
> > Meanwhile, it's still impossible to use the index in Info manuals on Windows,
> > and it's impossible to use some manuals (e.g. Viper) at all.
> 
> Don't worry, this will get fixed before Emacs 23 is ready for release.
> 
> I will work on it soon if no one beats me to it.

(Fore some value of "soon", sorry.)

I fixed this bug.

There were two suggestions for how to fix this: one by Handa-san in
this message:

  http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00293.html

followed by a tentative patch by Juanma here:

  http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00316.html

The other suggestion was by Stefan:

  http://lists.gnu.org/archive/html/emacs-devel/2008-08/msg00373.html

I decided I liked the first alternative better, since it has much
more local effect than the other one.  What Stefan suggested implied
messing with coding priorities, and I didn't feel that was TRT at this
late stage in Emacs 23.1 development.

The changes I installed are more thorough than what Juanma posted:
they change both detect_coding and detect_coding_system, and also bind
inhibit-null-byte-detection in a couple more places in info.el.

The result was tested on GNU/Linux, MS-Windows, and MS-DOS, both with
compressed and uncompressed Info files, with auto-compression-mode
both on and off.

Re: ^M in the info files

[Eli Zaretskii]

> Date: Sat, 10 Jan 2009 13:15:51 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-pretest-bug@gnu.org, bug-gnu-emacs@gnu.org, emacs-devel@gnu.org
> 
> The changes I installed are more thorough than what Juanma posted:
> they change both detect_coding and detect_coding_system, and also bind
> inhibit-null-byte-detection in a couple more places in info.el.

When documenting the new option, I found a strange inconsistency
between detect-coding-region/string and coding detection by
insert-file-contents.  If null bytes are detected, detect_coding sets
up to use no-conversion, but detect-coding-region/string does not call
detect_coding.  Instead, detect-coding-region/string call
detect_coding_system, which does not return no-conversion for a region
or string that include null bytes.  insert-file-contents does use
no-conversion for files that contain null bytes, but it does so
because decode_coding_gap does call detect_coding.

This creates an inconsistency for Lisp programs that do their own
decoding: if they call detect-coding-region/string for a region or
string with null bytes, they will not see no-conversion in the return
value, but insert-file-contents will use no-conversion nonetheless.

I think this inconsistency constitutes a bug.

Re: ^M in the info files

[Juanma Barranquero]

On Sat, Jan 10, 2009 at 12:15, Eli Zaretskii <eliz@gnu.org> wrote:

> I fixed this bug.

If the info files are compressed with gzip (using the Windows binary
from http://www.gzip.org), when visited from Info they still have
spurious ^M.

As an aside, when I use MSYS' gzip I get an error in decompressing:

  Error while executing "gzip -c -q -d < c:/emacs/info/efaq.gz"
  gzip: stdin: invalid compressed data--crc error

Could be related to CRLF issues in the redirection (but this is a gzip
problem, not Emacs', of course).

    Juanma

Re: ^M in the info files

[Eli Zaretskii]

> Date: Sat, 10 Jan 2009 15:07:37 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: 876-done@emacsbugs.donarmstrong.com, emacs-devel@gnu.org
> 
> On Sat, Jan 10, 2009 at 12:15, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > I fixed this bug.
> 
> If the info files are compressed with gzip (using the Windows binary
> from http://www.gzip.org), when visited from Info they still have
> spurious ^M.

I cannot reproduce this (tried with compressing info/emacs*).  Please
send a complete self-contained recipe.

Does "spurious" mean that every line has a ^M, like you saw before the
fix, or just some lines?

Also, does the above reference to http://www.gzip.org means that files
compressed by other versions of gzip do work as intended?

> As an aside, when I use MSYS' gzip I get an error in decompressing:
> 
>   Error while executing "gzip -c -q -d < c:/emacs/info/efaq.gz"
>   gzip: stdin: invalid compressed data--crc error
> 
> Could be related to CRLF issues in the redirection (but this is a gzip
> problem, not Emacs', of course).

Right, and I don't have MSYS anyway.

RE: ^M in the info files

[Drew Adams]

Thanks for fixing this. 

Re: ^M in the info files

[Juanma Barranquero]

On Sat, Jan 10, 2009 at 16:13, Eli Zaretskii <eliz@gnu.org> wrote:

> I cannot reproduce this (tried with compressing info/emacs*).  Please
> send a complete self-contained recipe.

Sorry, apparently it depends on something in my .emacs. It's still a
bug, though.

Try with

  cd info
  gzip efaq
  emacs -Q -eval "(progn (set-language-environment \"UTF-8\") (info
\"(efaq)Top\"))"

> Does "spurious" mean that every line has a ^M, like you saw before the
> fix, or just some lines?

Every line, just like before the fix.

> Also, does the above reference to http://www.gzip.org means that files
> compressed by other versions of gzip do work as intended?

Means that using the gzip from www.gzip.org it fails, and using the
one from MSYS I get the error I reported. I don't have other versions
of gzip.

For additional fun,

  cd info
  gzip emacs
  emacs -Q -eval "(info \"(emacs)Top\")"
   =>
  gzip: illegal option -- Q
  usage: gzip [-acdfhlLnNrtvV19] [-S suffix] [file ...

(this with the standard www.gzip.org version).

    Juanma

Re: ^M in the info files

[Eli Zaretskii]

Re: ^M in the info files

[Eli Zaretskii]

> Date: Sat, 10 Jan 2009 19:23:28 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: 876-done@emacsbugs.donarmstrong.com, emacs-devel@gnu.org
> 
> On Sat, Jan 10, 2009 at 16:13, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > I cannot reproduce this (tried with compressing info/emacs*).  Please
> > send a complete self-contained recipe.
> 
> Sorry, apparently it depends on something in my .emacs. It's still a
> bug, though.
> 
> Try with
> 
>   cd info
>   gzip efaq
>   emacs -Q -eval "(progn (set-language-environment \"UTF-8\") (info \"(efaq)Top\"))"

Yes, I see that now, but it has something to do with UTF-8 as the
language environment (it works correctly without it).  The Info buffer
has utf-8-unix as its buffer-file-coding-system, which is another
sign.  So I think this is a separate bug.  Please file a separate bug
report for it.

Re: ^M in the info files

[Eli Zaretskii]

> Date: Sat, 10 Jan 2009 19:23:28 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: 876-done@emacsbugs.donarmstrong.com, emacs-devel@gnu.org
> 
>   cd info
>   gzip emacs
>   emacs -Q -eval "(info \"(emacs)Top\")"
>    =>
>   gzip: illegal option -- Q
>   usage: gzip [-acdfhlLnNrtvV19] [-S suffix] [file ...
> 
> (this with the standard www.gzip.org version).

Doesn't happen to me, neither with gzip 1.2.4 from gzip.org, nor with
v1.3.5 from GnuWin32.  I did replace "emacs" with a full file name,
though, to make sure it doesn't pick up some other version somewhere
on my INFOPATH.

Re: ^M in the info files

[Juanma Barranquero]

On Sat, Jan 10, 2009 at 20:19, Eli Zaretskii <eliz@gnu.org> wrote:

>>   gzip: illegal option -- Q
>>   usage: gzip [-acdfhlLnNrtvV19] [-S suffix] [file ...
>>
>> (this with the standard www.gzip.org version).

My mistake. My command interpreter was trying to execute emacs.gz.
Sorry for the noise.

    Juanma

Re: ^M in the info files

[Lennart Borgman]

On Sat, Jan 10, 2009 at 5:21 PM, Drew Adams <drew.adams@oracle.com> wrote:
> Thanks for fixing this.

Yes, great Eli!

Re: ^M in the info files

[dhruva]

Hello,

On Sun, Jan 11, 2009 at 4:46 AM, Lennart Borgman
<lennart.borgman@gmail.com> wrote:
> On Sat, Jan 10, 2009 at 5:21 PM, Drew Adams <drew.adams@oracle.com> wrote:
>> Thanks for fixing this.
>
> Yes, great Eli!

I start emacs 'emacs -q' (built using MinGW on WXP, bzr HEAD) and see
^M in ada-mode. Do I need to 'bootstrap' or should a normal 'make all
install'.

-dhruva

-- 
Contents reflect my personal views only!

Re: ^M in the info files

[dhruva]

Hello,

On Sun, Jan 11, 2009 at 4:46 AM, Lennart Borgman
<lennart.borgman@gmail.com> wrote:
> On Sat, Jan 10, 2009 at 5:21 PM, Drew Adams <drew.adams@oracle.com> wrote:
>> Thanks for fixing this.
>
> Yes, great Eli!

I had to do a 'make clean' and 'make all install' (on WXP using MinGW,
bzr HEAD) to get it working. Thanks for the fix!

-dhruva

-- 
Contents reflect my personal views only!

Re: ^M in the info files

[Stefan Monnier]

Thank you, Eli,

> I decided I liked the first alternative better, since it has much
> more local effect than the other one.  What Stefan suggested implied
> messing with coding priorities, and I didn't feel that was TRT at this
> late stage in Emacs 23.1 development.

[ Obviously my suggestion requires more work to implement, but.. ]
Hmm...curious, why does it involve coding priorities?


        Stefan

Re: ^M in the info files

[Stefan Monnier]

> When documenting the new option, I found a strange inconsistency
> between detect-coding-region/string and coding detection by
> insert-file-contents.  If null bytes are detected, detect_coding sets
> up to use no-conversion, but detect-coding-region/string does not call
> detect_coding.  Instead, detect-coding-region/string call
> detect_coding_system, which does not return no-conversion for a region
> or string that include null bytes.  insert-file-contents does use
> no-conversion for files that contain null bytes, but it does so
> because decode_coding_gap does call detect_coding.

> This creates an inconsistency for Lisp programs that do their own
> decoding: if they call detect-coding-region/string for a region or
> string with null bytes, they will not see no-conversion in the return
> value, but insert-file-contents will use no-conversion nonetheless.

> I think this inconsistency constitutes a bug.

Indeed, it sounds like a bug.  Not sure how/when it would manifest
itself, tho.


        Stefan

Re: ^M in the info files

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: Kenichi Handa <handa@m17n.org>,  emacs-devel@gnu.org
> Date: Mon, 12 Jan 2009 15:56:16 -0500
> 
> > This creates an inconsistency for Lisp programs that do their own
> > decoding: if they call detect-coding-region/string for a region or
> > string with null bytes, they will not see no-conversion in the return
> > value, but insert-file-contents will use no-conversion nonetheless.
> 
> > I think this inconsistency constitutes a bug.
> 
> Indeed, it sounds like a bug.  Not sure how/when it would manifest
> itself, tho.

Well, for starters, it means that insert-file-contents does its own
detection that cannot be separated from it on the Lisp level.  That
is,

   (let ((coding-system-for-read 'undecided))
     (insert-file-contents FOO))

and

   (insert-file-contents-literally FOO)
   (let ((coding-system-for-read
          (detect-coding-region (point-min) (point-max) t)))
     (insert-file-contents FOO))

will surprisingly produce different results.

Re: ^M in the info files

[Eli Zaretskii]

> X-Spam-Status: No, score=-1.2 required=5.0 tests=AWL,BAYES_00,
> 	FORGED_RCVD_HELO,SPF_SOFTFAIL autolearn=no version=3.1.0
> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: 876-done@emacsbugs.donarmstrong.com,  emacs-pretest-bug@gnu.org,  bug-gnu-emacs@gnu.org,  emacs-devel@gnu.org
> Date: Mon, 12 Jan 2009 15:54:38 -0500
> 
> Hmm...curious, why does it involve coding priorities?

You suggested to introduce a way of controlling which possible
encodings will _not_ be acceptable by the caller of
detect-coding-region/string.  These primitives work by scanning the
coding_priorities[] array and checking them against those categories
that has been rejected based on the text in the region/string.  See
detect_coding and detect_coding_system.

Re: ^M in the info files

[Glenn Morris]

Please stop cc'ing emacs-pretest-bug@gnu.org and bug-gnu-emacs@gnu.org
in this discussion, it messes up the bug tracker.

Re: ^M in the info files

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Cc: Stefan Monnier <monnier@iro.umontreal.ca>, 876@emacsbugs.donarmstrong.com,  emacs-devel@gnu.org
> Date: Mon, 12 Jan 2009 17:27:03 -0500
> 
> 
> Please stop cc'ing emacs-pretest-bug@gnu.org and bug-gnu-emacs@gnu.org
> in this discussion, it messes up the bug tracker.

Why won't the tracker get its act together and stop forcing us into
editing mail headers?  I use the Rmail's `r' command, which replies to
all the original "To" and "CC" addressees.  With the amounts of mail
I'm responding to, I cannot afford editing headers of every message I
send.  Even if I try, I will surely sometimes forget.

bug bugs [was Re: ^M in the info files]

[Glenn Morris]

Eli Zaretskii wrote:

> Why won't the tracker get its act together and stop forcing us into
> editing mail headers?

It's been asked for repeatedly, but it doesn't happen. There seems no
point even discussing it, all the points have been made.

I know people shouldn't have to make this effort with headers, but I
was only asking. I'm fed up of duplicate and triplicate mails, and
large numbers of bogus bug reports that need merging.

>  I use the Rmail's `r' command, which replies to all the original
> "To" and "CC" addressees. 

I don't use rmail, but rmail-dont-reply-to-names ...?

Re: bug bugs [was Re: ^M in the info files]

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Cc: monnier@iro.umontreal.ca,  emacs-devel@gnu.org
> Date: Mon, 12 Jan 2009 23:24:35 -0500
> 
> rmail-dont-reply-to-names ...?

Sure, but I cannot use rmail-dont-reply-to-names in this case, I
think, since it means I will never reply to these mailing lists, even
if the message didn't come from the bug tracker.  Those lists aren't
yet usurped by the bug tracker, are they?  People can still post there
without their messages being taken as bug reports, can't they?

Re: bug bugs

[Glenn Morris]

Eli Zaretskii wrote:

> Those lists aren't yet usurped by the bug tracker, are they?

Yes they are. (There is no longer "lists" plural either, they are
identical.)

> People can still post there without their messages being taken as
> bug reports, can't they?

No. (Well, except for bug#936, the broken newsgroup gateway.)

Re: bug bugs

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Cc: monnier@iro.umontreal.ca,  emacs-devel@gnu.org
> Date: Tue, 13 Jan 2009 15:50:16 -0500
> 
> Eli Zaretskii wrote:
> 
> > Those lists aren't yet usurped by the bug tracker, are they?
> 
> Yes they are.

So where do humans have chance of talking about bugs?

Re: bug bugs

[Chong Yidong]

Eli Zaretskii <eliz@gnu.org> writes:

> So where do humans have chance of talking about bugs?

Send a CC to NNNN@emacsbugs.donarmstrong.com, where NNNN is the bug
number, instead of bug-gnu-emacs@gnu.org; these will show up on the
mailing list too.

Re: bug bugs

[Eli Zaretskii]

> From: Chong Yidong <cyd@stupidchicken.com>
> Cc: Glenn Morris <rgm@gnu.org>,  monnier@iro.umontreal.ca,  emacs-devel@gnu.org
> Date: Wed, 14 Jan 2009 08:45:28 -0500
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > So where do humans have chance of talking about bugs?
> 
> Send a CC to NNNN@emacsbugs.donarmstrong.com, where NNNN is the bug
> number, instead of bug-gnu-emacs@gnu.org; these will show up on the
> mailing list too.

Are you saying that if I want to talk to you about bugs, I need to do
that via the bug tracker?  I find that a bit silly.

Re: bug bugs

[Glenn Morris]

Eli Zaretskii wrote:

> So where do humans have chance of talking about bugs?

That's somewhat off my topic. I simply want to assure you that the the
_only_ reason to send mail to bug-gnu-emacs or emacs-pretest-bug is to
create a new bug in the tracker. It's been that way for months and
months. Those addresses should never be appearing in replies. At best,
it creates pointless duplicate mails; at worst it creates new,
separate bug reports that need to be tidied up.

Re: bug bugs

[Stephen J. Turnbull]

Eli Zaretskii writes:

 > Are you saying that if I want to talk to you about bugs, I need to do
 > that via the bug tracker?  I find that a bit silly.

That's not "silly", that's "process".  The intent is that useful
information be located where it is most likely to be used, and found
in the future.