"coding: utf-8" emacs local variable in ChangeLog (?)

"coding: utf-8" emacs local variable in ChangeLog (?)

[Joel Brobecker]

[-- Attachment #1: Type: text/plain, Size: 558 bytes --]

Hello,

While rotating the ChangeLog file, I noticed a discrepancy between
what the documentation says we should add at the end of the file
for emacs:

    Local Variables:
    mode: change-log
    left-margin: 8
    fill-column: 74
    version-control: never
    End:

And what we actually have. Right now, we have an extra line:

    +coding: utf-8

I tracked the change to a patch in Aug 2007, but couldn't find
the associated discussion in the gdb and gdb-patches mailing list.

Is that change OK? If yes, an update of gdbint.texi is in order.

-- 
Joel

[-- Attachment #2: coding.diff --]
[-- Type: text/plain, Size: 999 bytes --]

commit bfbc00954f898246d7d3b8aa51bba9136acfaf9d
Author: Ludovic CourtÚs <ludo@gnu.org>
Date:   Thu Aug 9 22:44:38 2007 +0000

    Add Emacs local variables for UTF-8.

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index eedb1cc..914fe17 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,8 @@
+2007-08-10  Ludovic CourtÚs  <ludo@gnu.org>
+
+	* MAINTAINERS: Add Emacs local variables to use UTF-8 upon
+	opening.
+
 2007-08-09  Ludovic CourtÚs  <ludo@gnu.org>
 
 	* MAINTAINERS (Write After Approval): Add myself.
@@ -6236,4 +6241,5 @@ mode: change-log
 left-margin: 8
 fill-column: 74
 version-control: never
+coding: utf-8
 End:
diff --git a/gdb/MAINTAINERS b/gdb/MAINTAINERS
index a70cd6c..982e60b 100644
--- a/gdb/MAINTAINERS
+++ b/gdb/MAINTAINERS
@@ -617,3 +617,7 @@ Michael Chastain (testsuite)			mec.gnu at mindspring dot com
 Folks that have been caught up in a paper trail:
 
 David Carlton					carlton@bactrian.org
+
+;; Local Variables:
+;; coding: utf-8
+;; End:

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Sat, 3 Jan 2009 10:13:25 +0400
> From: Joel Brobecker <brobecker@adacore.com>
> Cc: ludo@gnu.org
> 
>     +coding: utf-8
> 
> I tracked the change to a patch in Aug 2007, but couldn't find
> the associated discussion in the gdb and gdb-patches mailing list.
> 
> Is that change OK?

Yes.

> If yes, an update of gdbint.texi is in order.

Please do, and thanks.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Mark Kettenis]

> Date: Sat, 03 Jan 2009 11:48:00 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Sat, 3 Jan 2009 10:13:25 +0400
> > From: Joel Brobecker <brobecker@adacore.com>
> > Cc: ludo@gnu.org
> > 
> >     +coding: utf-8
> > 
> > I tracked the change to a patch in Aug 2007, but couldn't find
> > the associated discussion in the gdb and gdb-patches mailing list.
> > 
> > Is that change OK?
> 
> Yes.

Well, I disagree.  I think it was a mistake.  UTF-8 support is not
universally available.  I can see why Ludovic wants to have his name
spelled properly, but he should realize it'll look like "Ludovic
Court?\203?is" to many people.  I think we should keep our source
files as 7-bit ASCII (with the possible exception of the files
maintained by the translation teams), Developers should use a
transliteration for their names, just like all deveopers from
countries in east-asia or the former (current?) Russian empire seem to
do.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Ludovic Courtès]

Hello,

Joel Brobecker <brobecker@adacore.com> writes:

> Is that change OK? If yes, an update of gdbint.texi is in order.

FWIW, my name in the Git commit is misspelled.

Mark Kettenis <mark.kettenis@xs4all.nl> writes:

> Well, I disagree.  I think it was a mistake.  UTF-8 support is not
> universally available.  I can see why Ludovic wants to have his name
> spelled properly, but he should realize it'll look like "Ludovic
> Court?\203?is" to many people.

Actually I think it would just look weird to relatively few people (most
GNU/Linux distros come with UTF-8 terminals by default, for instance)
but would look OK to others.  Transliteration ought to be avoided as
much as possible IMO.

Thanks,
Ludo'.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Sat, 3 Jan 2009 11:22:53 +0100 (CET)
> From: Mark Kettenis <mark.kettenis@xs4all.nl>
> CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> 
> Well, I disagree.  I think it was a mistake.  UTF-8 support is not
> universally available.

Any specific examples?

> I can see why Ludovic wants to have his name spelled properly, but
> he should realize it'll look like "Ludovic Court?\203?is" to many
> people.

Not if they use Emacs.

> I think we should keep our source files as 7-bit ASCII (with the
> possible exception of the files maintained by the translation
> teams), Developers should use a transliteration for their names,
> just like all deveopers from countries in east-asia or the former
> (current?) Russian empire seem to do.

I doubt if you can get that agreed to by many people who live in the
Latin locales.  (But if you succeed, I won't mind going back to ASCII,
of course.)

Also note that leaving the encoding unspecified will almost always
produce incorrect display if someone uses non-ASCII characters, while
specifying the encoding at least does TRT in Emacs.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Mark Kettenis]

> Date: Sat, 03 Jan 2009 14:45:18 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Sat, 3 Jan 2009 11:22:53 +0100 (CET)
> > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > 
> > Well, I disagree.  I think it was a mistake.  UTF-8 support is not
> > universally available.
> 
> Any specific examples?

Sure, the OpenBSD system console defenitely isn't UTF-8 aware, and as
far as I can tell the xterm's on OpenBSD don't have UTF-8 encoding
turned on by default either.

> > I can see why Ludovic wants to have his name spelled properly, but
> > he should realize it'll look like "Ludovic Court?\203?is" to many
> > people.
> 
> Not if they use Emacs.

What do you think I'm using?  Ok, that's in emacs running in screen
running in an xterm, and those question marks are really some funny
looking character that I wouldn't know how to type.  But even if I
look at the 2007 ChangeLog with emacs running as an X application
Ludo's last name shows up with that funny capital A with a tiled and "
after it.  I'm fairly sure that's not right.

> > I think we should keep our source files as 7-bit ASCII (with the
> > possible exception of the files maintained by the translation
> > teams), Developers should use a transliteration for their names,
> > just like all deveopers from countries in east-asia or the former
> > (current?) Russian empire seem to do.
> 
> I doubt if you can get that agreed to by many people who live in the
> Latin locales.  (But if you succeed, I won't mind going back to ASCII,
> of course.)

Well, their ancestors didn't have them ;)

Seriously though.  This is not worth fighting about.  I stated my
opinion.  I'll get back under my rock.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Sun, 4 Jan 2009 00:56:28 +0100 (CET)
> From: Mark Kettenis <mark.kettenis@xs4all.nl>
> CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> 
> > > I can see why Ludovic wants to have his name spelled properly, but
> > > he should realize it'll look like "Ludovic Court?\203?is" to many
> > > people.
> > 
> > Not if they use Emacs.
> 
> What do you think I'm using?  Ok, that's in emacs running in screen
> running in an xterm, and those question marks are really some funny
> looking character that I wouldn't know how to type.

"C-u C-x =" will spell what is that character, and there's
latin1-disp.el that can display quite a few of these characters even
on a text terminal.

> But even if I look at the 2007 ChangeLog with emacs running as an X
> application Ludo's last name shows up with that funny capital A with
> a tiled and " after it.  I'm fairly sure that's not right.

That's _exactly_ the reason to add the `coding' cookie: you were
looking at what happens when Emacs tries to interpret a utf-8 encoded
text as if it were Latin-1 (or maybe Latin-9) encoded.  Now, after the
cookie was added, I see Ludo's name displayed correctly.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Mark Kettenis]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1852 bytes --]

> Date: Tue, 06 Jan 2009 22:34:22 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Sun, 4 Jan 2009 00:56:28 +0100 (CET)
> > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > 
> > > > I can see why Ludovic wants to have his name spelled properly, but
> > > > he should realize it'll look like "Ludovic Court?\203?is" to many
> > > > people.
> > > 
> > > Not if they use Emacs.
> > 
> > What do you think I'm using?  Ok, that's in emacs running in screen
> > running in an xterm, and those question marks are really some funny
> > looking character that I wouldn't know how to type.
> 
> "C-u C-x =" will spell what is that character, and there's
> latin1-disp.el that can display quite a few of these characters even
> on a text terminal.

This is what C-u C-x says about the funny character in Ludo's name:

    character: Ã (0303, 195, 0xc3)
      charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))
   code point: 195
       syntax: whitespace
     category:
  buffer code: 0xC3
    file code: 0xC3 (encoded by coding system raw-text-unix)
terminal code: not encodable

> > But even if I look at the 2007 ChangeLog with emacs running as an X
> > application Ludo's last name shows up with that funny capital A with
> > a tiled and " after it.  I'm fairly sure that's not right.
> 
> That's _exactly_ the reason to add the `coding' cookie: you were
> looking at what happens when Emacs tries to interpret a utf-8 encoded
> text as if it were Latin-1 (or maybe Latin-9) encoded.  Now, after the
> cookie was added, I see Ludo's name displayed correctly.

I'm not saying that adding the `coding' cookie is a problem.  I'm
arguing that turning ChangeLog into a UTF-8 encoded file was a bad
decision, since there are enough systems where it will not be properly
displayed.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Tue, 6 Jan 2009 22:25:36 +0100 (CET)
> From: Mark Kettenis <mark.kettenis@xs4all.nl>
> CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> 
> This is what C-u C-x says about the funny character in Ludo's name:
> 
>     character: Ã (0303, 195, 0xc3)
>       charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))

That's again because your defaults are set to interpret this as
something other than UTF-8.

> I'm not saying that adding the `coding' cookie is a problem.  I'm
> arguing that turning ChangeLog into a UTF-8 encoded file was a bad
> decision, since there are enough systems where it will not be properly
> displayed.

And I'm arguing that adding the coding cookie will cause this to be
displayed properly in Emacs.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Mark Kettenis]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1047 bytes --]

> Date: Tue, 06 Jan 2009 23:59:29 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Tue, 6 Jan 2009 22:25:36 +0100 (CET)
> > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > 
> > This is what C-u C-x says about the funny character in Ludo's name:
> > 
> >     character: Ã (0303, 195, 0xc3)
> >       charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))
> 
> That's again because your defaults are set to interpret this as
> something other than UTF-8.

They're not.  This is on a standard emacs 21.4 and there is *nothing*
in my .emacs that doesn't touch anything related to 'coding'.

> > I'm not saying that adding the `coding' cookie is a problem.  I'm
> > arguing that turning ChangeLog into a UTF-8 encoded file was a bad
> > decision, since there are enough systems where it will not be properly
> > displayed.
> 
> And I'm arguing that adding the coding cookie will cause this to be
> displayed properly in Emacs.

Obviously that is not true on all systems.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Tue, 6 Jan 2009 23:46:58 +0100 (CET)
> From: Mark Kettenis <mark.kettenis@xs4all.nl>
> CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> 
> > Date: Tue, 06 Jan 2009 23:59:29 +0200
> > From: Eli Zaretskii <eliz@gnu.org>
> > 
> > > Date: Tue, 6 Jan 2009 22:25:36 +0100 (CET)
> > > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > > 
> > > This is what C-u C-x says about the funny character in Ludo's name:
> > > 
> > >     character: Ã (0303, 195, 0xc3)
> > >       charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))
> > 
> > That's again because your defaults are set to interpret this as
> > something other than UTF-8.
> 
> They're not.  This is on a standard emacs 21.4 and there is *nothing*
> in my .emacs that doesn't touch anything related to 'coding'.

In which case Emacs uses your locale to set up things.  I'm guessing
that your locale is not a UTF-8 one, probably Latin-1 or Latin-9?

> > > I'm not saying that adding the `coding' cookie is a problem.  I'm
> > > arguing that turning ChangeLog into a UTF-8 encoded file was a bad
> > > decision, since there are enough systems where it will not be properly
> > > displayed.
> > 
> > And I'm arguing that adding the coding cookie will cause this to be
> > displayed properly in Emacs.
> 
> Obviously that is not true on all systems.

"Obviously"?  You mean, Ludo's name is not displayed correctly on your
system even after the `coding' cookie was added?

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Mark Kettenis]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1835 bytes --]

> Date: Wed, 07 Jan 2009 06:18:55 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Tue, 6 Jan 2009 23:46:58 +0100 (CET)
> > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > 
> > > Date: Tue, 06 Jan 2009 23:59:29 +0200
> > > From: Eli Zaretskii <eliz@gnu.org>
> > > 
> > > > Date: Tue, 6 Jan 2009 22:25:36 +0100 (CET)
> > > > From: Mark Kettenis <mark.kettenis@xs4all.nl>
> > > > CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> > > > 
> > > > This is what C-u C-x says about the funny character in Ludo's name:
> > > > 
> > > >     character: Ã (0303, 195, 0xc3)
> > > >       charset: eight-bit-graphic (8-bit graphic char (0xA0..0xFF))
> > > 
> > > That's again because your defaults are set to interpret this as
> > > something other than UTF-8.
> > 
> > They're not.  This is on a standard emacs 21.4 and there is *nothing*
> > in my .emacs that doesn't touch anything related to 'coding'.
> 
> In which case Emacs uses your locale to set up things.  I'm guessing
> that your locale is not a UTF-8 one, probably Latin-1 or Latin-9?

I'm using the locale that <deity> intended to be used on a UNIX
system: the "C" locale ;).

> > > > I'm not saying that adding the `coding' cookie is a problem.
> > > > I'm arguing that turning ChangeLog into a UTF-8 encoded file
> > > > was a bad decision, since there are enough systems where it
> > > > will not be properly displayed.
> > > 
> > > And I'm arguing that adding the coding cookie will cause this to be
> > > displayed properly in Emacs.
> > 
> > Obviously that is not true on all systems.
> 
> "Obviously"?  You mean, Ludo's name is not displayed correctly on your
> system even after the `coding' cookie was added?

Yes, that's what I have been trying to tell you in the last couple of mails.

Re: "coding: utf-8" emacs local variable in ChangeLog (?)

[Eli Zaretskii]

> Date: Wed, 7 Jan 2009 11:05:52 +0100 (CET)
> From: Mark Kettenis <mark.kettenis@xs4all.nl>
> CC: brobecker@adacore.com, gdb@sourceware.org, ludo@gnu.org
> 
> > In which case Emacs uses your locale to set up things.  I'm guessing
> > that your locale is not a UTF-8 one, probably Latin-1 or Latin-9?
> 
> I'm using the locale that <deity> intended to be used on a UNIX
> system: the "C" locale ;).

The "C" locale has Latin-1 as its default for non-ASCII, IIRC.  So
that's why Emacs tries to decode the file as Latin-1, when there's no
`coding' cookie.

> > > > > I'm not saying that adding the `coding' cookie is a problem.
> > > > > I'm arguing that turning ChangeLog into a UTF-8 encoded file
> > > > > was a bad decision, since there are enough systems where it
> > > > > will not be properly displayed.
> > > > 
> > > > And I'm arguing that adding the coding cookie will cause this to be
> > > > displayed properly in Emacs.
> > > 
> > > Obviously that is not true on all systems.
> > 
> > "Obviously"?  You mean, Ludo's name is not displayed correctly on your
> > system even after the `coding' cookie was added?
> 
> Yes, that's what I have been trying to tell you in the last couple of mails.

Sorry, I must have missed that.  Please tell again: is it in the GUI
session or on a text-only terminal?