bug#18205: Obsolete patterns in auto-mode-alist

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 956 bytes --]

The default value for auto-mode-alist currently contains the following:

     ;; .emacs or .gnus or .viper following a directory delimiter in
     ;; Unix, MSDOG or VMS syntax.
     ("[]>:/\\]\\..*\\(emacs\\|gnus\\|viper\\)\\'" . emacs-lisp-mode)
     ("\\`\\..*emacs\\'" . emacs-lisp-mode)
     ;; _emacs following a directory delimiter
     ;; in MsDos syntax
     ("[:/]_emacs\\'" . emacs-lisp-mode)

There are two non-standard spellings:

s/MSDOG/MS-DOS/
s/MsDos/MS-DOS/

Further, the second line looks like it will only match a .emacs in the root
directory, presumably because the first line will not.

Given that VMS support has been removed, can we replace the first two
pattern lines with a line that uses the pattern used above:

\\(/\\|\\`\\)

to match either start of line or directory separator? And can we then also
remove the third line, as MS-DOS syntax is not treated specially anywhere
else in this default value?

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1267 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Stefan Monnier]

> There are two non-standard spellings:

> s/MSDOG/MS-DOS/

This used to be used as a joke in Emacs source.  Nowadays we should use
the name of a Free Software instead.

FreeDOS maybe?


        Stefan

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 521 bytes --]

On 6 August 2014 18:41, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> > There are two non-standard spellings:
>
> > s/MSDOG/MS-DOS/
>
> This used to be used as a joke in Emacs source.  Nowadays we should use
> the name of a Free Software instead.
>
> FreeDOS maybe?
>

That would be confusing (since it's still best known as MS-DOS) and
inconsistent (since Emacs calls it MS-DOS or ms-dos everywhere). But if my
suggestion about the patterns is taken, these comments will disappear
anyway.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 942 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Wed, 6 Aug 2014 18:47:14 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: 18205@debbugs.gnu.org
> 
> > > s/MSDOG/MS-DOS/
> >
> > This used to be used as a joke in Emacs source.  Nowadays we should use
> > the name of a Free Software instead.
> >
> > FreeDOS maybe?
> >
> 
> That would be confusing (since it's still best known as MS-DOS) and
> inconsistent (since Emacs calls it MS-DOS or ms-dos everywhere).

Did you search the Emacs tree for MS-DOG?

bug#18205: Obsolete patterns in auto-mode-alist

[Glenn Morris]

Stefan Monnier wrote:

> Nowadays we should use the name of a Free Software instead.
>
> FreeDOS maybe?

This would be nice, but IIUC no-one has ever tested Emacs under FreeDOS.
It's DJGPP MS-DOS only.

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Date: Wed, 06 Aug 2014 20:14:43 -0400
> Cc: 18205@debbugs.gnu.org, Reuben Thomas <rrt@sc3d.org>
> 
> Stefan Monnier wrote:
> 
> > Nowadays we should use the name of a Free Software instead.
> >
> > FreeDOS maybe?
> 
> This would be nice, but IIUC no-one has ever tested Emacs under FreeDOS.
> It's DJGPP MS-DOS only.

DJGPP supports FreeDOS.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 627 bytes --]

On 7 August 2014 03:53, Eli Zaretskii <eliz@gnu.org> wrote:

> > From: Glenn Morris <rgm@gnu.org>
> > Date: Wed, 06 Aug 2014 20:14:43 -0400
> > Cc: 18205@debbugs.gnu.org, Reuben Thomas <rrt@sc3d.org>
> >
> > Stefan Monnier wrote:
> >
> > > Nowadays we should use the name of a Free Software instead.
> > >
> > > FreeDOS maybe?
> >
> > This would be nice, but IIUC no-one has ever tested Emacs under FreeDOS.
> > It's DJGPP MS-DOS only.
>
> DJGPP supports FreeDOS.
>

In that case, perhaps it would be better to refer to DJGPP everywhere, so
then we'd be referring to free software instead of non-free?

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1282 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 837 bytes --]

On 7 August 2014 10:39, Reuben Thomas <rrt@sc3d.org> wrote:

>
> On 7 August 2014 03:53, Eli Zaretskii <eliz@gnu.org> wrote:
>
>> > From: Glenn Morris <rgm@gnu.org>
>> > Date: Wed, 06 Aug 2014 20:14:43 -0400
>> > Cc: 18205@debbugs.gnu.org, Reuben Thomas <rrt@sc3d.org>
>> >
>> > Stefan Monnier wrote:
>> >
>> > > Nowadays we should use the name of a Free Software instead.
>> > >
>> > > FreeDOS maybe?
>> >
>> > This would be nice, but IIUC no-one has ever tested Emacs under FreeDOS.
>> > It's DJGPP MS-DOS only.
>>
>> DJGPP supports FreeDOS.
>>
>
> In that case, perhaps it would be better to refer to DJGPP everywhere, so
> then we'd be referring to free software instead of non-free?
>

Either way, I've completed the rename (r117661), so that MS-DOS is now
referred to by the same name everywhere (AFAICT).

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1910 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 975 bytes --]

On 7 August 2014 11:07, Reuben Thomas <rrt@sc3d.org> wrote:

> On 7 August 2014 10:39, Reuben Thomas <rrt@sc3d.org> wrote:
>
>>
>> On 7 August 2014 03:53, Eli Zaretskii <eliz@gnu.org> wrote:
>>
>>> > From: Glenn Morris <rgm@gnu.org>
>>> > Date: Wed, 06 Aug 2014 20:14:43 -0400
>>> > Cc: 18205@debbugs.gnu.org, Reuben Thomas <rrt@sc3d.org>
>>> >
>>> > Stefan Monnier wrote:
>>> >
>>> > > Nowadays we should use the name of a Free Software instead.
>>> > >
>>> > > FreeDOS maybe?
>>> >
>>> > This would be nice, but IIUC no-one has ever tested Emacs under
>>> FreeDOS.
>>> > It's DJGPP MS-DOS only.
>>>
>>> DJGPP supports FreeDOS.
>>>
>>
>> In that case, perhaps it would be better to refer to DJGPP everywhere, so
>> then we'd be referring to free software instead of non-free?
>>
>
> Either way, I've completed the rename (r117661), so that MS-DOS is now
> referred to by the same name everywhere (AFAICT).
>

In fact, I missed a few, hence r117662.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 2447 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

    Either way, I've completed the rename (r117661), so that MS-DOS is now
    referred to by the same name everywhere (AFAICT).

Did you use our name for the system, or Microsoft's name for it?
Uniformity in how we refer to that system is NOT necessary,
but I object to removing the name "MSDOG".

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call.

bug#18205: Obsolete patterns in auto-mode-alist

[Stefan Monnier]

> In that case, perhaps it would be better to refer to DJGPP everywhere, so
> then we'd be referring to free software instead of non-free?

DJGPP or FreeDOS is fine by me (or even just DOS).  I think Eli would
know better which is more technically appropriate, but I think we should
prefer references to Free Software products over references to
proprietary products (even if long dead).


        Stefan

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: Eli Zaretskii <eliz@gnu.org>,  Glenn Morris <rgm@gnu.org>,  18205@debbugs.gnu.org
> Date: Thu, 07 Aug 2014 08:23:58 -0400
> 
> > In that case, perhaps it would be better to refer to DJGPP everywhere, so
> > then we'd be referring to free software instead of non-free?
> 
> DJGPP or FreeDOS is fine by me (or even just DOS).

DJGPP is not a name of an OS, it's a development environment.  So it
cannot simply replace MS-DOS.

> I think Eli would know better which is more technically appropriate,
> but I think we should prefer references to Free Software products
> over references to proprietary products (even if long dead).

I think we already invested too much energy in discussing a non-issue.

bug#18205: Obsolete patterns in auto-mode-alist

[Stefan Monnier]

> I think we already invested too much energy in discussing a non-issue.

You do have a point about this discussion being silly, but I think the
core is that we don't want to advertize proprietary products, so in
order to do that while still being able to refer to such products, we
like to garble the product name in a way that expresses our dislike
for it (so it's clear we don't advocate its use).

"MS-DOG" and "Losedows" were examples of that.  I'm not very fond of
such tricks, but if we get rid of MS-DOG I think it's better to replace
it with FreeDOS than with MS-DOS.


        Stefan

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: rrt@sc3d.org,  rgm@gnu.org,  18205@debbugs.gnu.org
> Date: Thu, 07 Aug 2014 15:12:46 -0400
> 
> > I think we already invested too much energy in discussing a non-issue.
> 
> You do have a point about this discussion being silly, but I think the
> core is that we don't want to advertize proprietary products, so in
> order to do that while still being able to refer to such products, we
> like to garble the product name in a way that expresses our dislike
> for it (so it's clear we don't advocate its use).
> 
> "MS-DOG" and "Losedows" were examples of that.  I'm not very fond of
> such tricks, but if we get rid of MS-DOG I think it's better to replace
> it with FreeDOS than with MS-DOS.

I say let the sleeping dog lie.  The MS-DOS port of Emacs is in
maintenance mode, and even I build it and work on it only during
pretests.  It makes very little sense to "fix" anything related to it
that isn't 100% broken.  It certainly makes no sense to start yet
another argument over it with RMS.

bug#18205: Obsolete patterns in auto-mode-alist

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

Since MSDOG is a well-known operating system, among those who might
possibly use it, we have no reason to try to avoid mentioning it.
So there is no reason to try to avoid saying "MSDOG" where that system
is in question.

If the code or feature being documented works also with FreeDOS, we
could mention that too.  (I don't know whether FreeDOS is a 100% free
OS.)

It would not make sense to delete the reference to MSDOG and mention
only FreeDOS, because that would not give users the information we
want to give them.

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call.

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Thu, 07 Aug 2014 14:24:18 -0400
> From: Richard Stallman <rms@gnu.org>
> CC: rgm@gnu.org, 18205@debbugs.gnu.org, rrt@sc3d.org
> 
> I don't know whether FreeDOS is a 100% free OS.

Its license is GPL.

bug#18205: Obsolete patterns in auto-mode-alist

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

    > I don't know whether FreeDOS is a 100% free OS.

    Its license is GPL.

Is it complete and self-hosting?

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call.

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Fri, 08 Aug 2014 01:35:35 -0400
> From: Richard Stallman <rms@gnu.org>
> CC: rgm@gnu.org, 18205@debbugs.gnu.org, rrt@sc3d.org
> 
>     > I don't know whether FreeDOS is a 100% free OS.
> 
>     Its license is GPL.
> 
> Is it complete and self-hosting?

Yes.

bug#18205: Obsolete patterns in auto-mode-alist

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

Would you like to ask the developers of Freedos if they want
to have it listed as a non-GNU free distro?

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call.

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Fri, 08 Aug 2014 15:11:15 -0400
> From: Richard Stallman <rms@gnu.org>
> CC: rgm@gnu.org, 18205@debbugs.gnu.org, rrt@sc3d.org
> 
> Would you like to ask the developers of Freedos if they want
> to have it listed as a non-GNU free distro?

Sorry, I don't know who develops FreeDOS, and I never communicated
with any of them.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 388 bytes --]

On 8 August 2014 20:11, Richard Stallman <rms@gnu.org> wrote:

> Would you like to ask the developers of Freedos if they want
> to have it listed as a non-GNU free distro?
>

Sure. I can't find any such listing currently; is there a page I can point
them to show them what I'm talking about, or do I just say "a new page like
our list of free GNU/Linux distros"?

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 839 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 1710 bytes --]

On 8 August 2014 07:06, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Fri, 08 Aug 2014 01:35:35 -0400
> > From: Richard Stallman <rms@gnu.org>
> > CC: rgm@gnu.org, 18205@debbugs.gnu.org, rrt@sc3d.org
> >
> >     > I don't know whether FreeDOS is a 100% free OS.
> >
> >     Its license is GPL.
> >
> > Is it complete and self-hosting?
>
> Yes.
>

I spent a while investigating Emacs on FreeDOS. The results go way beyond
the scope of this bug report, so I'll file other bugs, or write to
gnu-prog-discuss as appropriate. In summary, though:

1. Emacs 23.3, as distributed for DJGPP, works fine on FreeDOS.

2. The MS-DOS build system was broken in Emacs 24.3, i.e. the most recent
release. (There are some problems that may be specific to FreeDOS, such as
an inability to understand "@call" in config.bat, but the lack of the
msdos/autogen/ directory in the tarball means that the build won't work; it
still doesn't work after copying that directory, though I think the
remaining problems are down to the lack of long file-name support in some
of the native FreeDOS utilities.)

3. Rather than try to repair the MS-DOS build system (which remains,
AFAICS, the only documented way to build Emacs for MS-DOS), it seems more
productive to get the normal build system working. Given that DJGPP
distributes a reasonably comprehensive and up-to-date set of GNU tools,
this shouldn't be too hard. I built a distribution of Bazaar head, and
after hacking it to accept the msdosdjgpp OS name, and a couple of other
small hacks that I'll write up, it gets all the way through the configure
script, crashing with an error I've not yet been able to diagnose when it
tries to write config.status.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 2604 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Sat, 9 Aug 2014 00:19:39 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>, 18205@debbugs.gnu.org
> 
> 2. The MS-DOS build system was broken in Emacs 24.3, i.e. the most recent
> release.

It works fine for me (but I don't have FreeDOS installed and didn't
try to build on FreeDOS).

> 3. Rather than try to repair the MS-DOS build system (which remains,
> AFAICS, the only documented way to build Emacs for MS-DOS), it seems more
> productive to get the normal build system working.

Feel free to do that, but personally, I think this is a waste of
energy.  From my POV, the DJGPP build of Emacs is in maintenance mode:
no new features specific to that build are or should be considered.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 484 bytes --]

On 9 August 2014 07:49, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Sat, 9 Aug 2014 00:19:39 +0100
> > From: Reuben Thomas <rrt@sc3d.org>
> > Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>,
> 18205@debbugs.gnu.org
> >
> > 2. The MS-DOS build system was broken in Emacs 24.3, i.e. the most recent
> > release.
>
> It works fine for me (but I don't have FreeDOS installed and didn't
> try to build on FreeDOS).
>

OK, I'll file a bug report.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1080 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 1056 bytes --]

On 9 August 2014 07:49, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Sat, 9 Aug 2014 00:19:39 +0100
> > From: Reuben Thomas <rrt@sc3d.org>
> > Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>,
> 18205@debbugs.gnu.org
> >
> > 2. The MS-DOS build system was broken in Emacs 24.3, i.e. the most recent
> > release.
>
> It works fine for me (but I don't have FreeDOS installed and didn't
> try to build on FreeDOS).
>
> > 3. Rather than try to repair the MS-DOS build system (which remains,
> > AFAICS, the only documented way to build Emacs for MS-DOS), it seems more
> > productive to get the normal build system working.
>
> Feel free to do that, but personally, I think this is a waste of
> energy.  From my POV, the DJGPP build of Emacs is in maintenance mode:
> no new features specific to that build are or should be considered.
>

That's pretty much the point: I am suggesting removing features specific to
the build (in this case, the build system) and letting the normal build
system take the strain instead.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1753 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Sat, 9 Aug 2014 08:44:46 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>, 18205@debbugs.gnu.org
> 
> > > 3. Rather than try to repair the MS-DOS build system (which remains,
> > > AFAICS, the only documented way to build Emacs for MS-DOS), it seems more
> > > productive to get the normal build system working.
> >
> > Feel free to do that, but personally, I think this is a waste of
> > energy.  From my POV, the DJGPP build of Emacs is in maintenance mode:
> > no new features specific to that build are or should be considered.
> >
> 
> That's pretty much the point: I am suggesting removing features specific to
> the build (in this case, the build system) and letting the normal build
> system take the strain instead.

I envision a non-trivial effort doing that, an effort that is IMO a
waste of valuable resources.

The current DJGPP build "just works", given that you have a correct
setup on a system that supports the build (requirements for supporting
the build might be different from those for supporting running of the
binary).  I have built all the pretests of Emacs 24.4 with no
problems.  I have reports from people who successfully built the
previous version, 24.3.  The problems you saw on FreeDOS using the
current build system are either due to FreeDOS specific issues or to
your particular setup and the tools you installed.  We could try
debugging those problems, but unless you intend to continue building
the DJGPP port, that, too, would be a waste, IMO.

Given the above, investing efforts in moving DJGPP to the Posix
configury is a waste of resources that are best applied to parts of
Emacs that will benefit users of modern systems.

Again, if after all I've said, you still want to work on that, feel
free to do it, and thanks in advance.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 1984 bytes --]

On 9 August 2014 09:30, Eli Zaretskii <eliz@gnu.org> wrote:

> > That's pretty much the point: I am suggesting removing features specific
> to
> > the build (in this case, the build system) and letting the normal build
> > system take the strain instead.
>
> I envision a non-trivial effort doing that, an effort that is IMO a
> waste of valuable resources.
>
> The current DJGPP build "just works", given that you have a correct
> setup on a system that supports the build (requirements for supporting
> the build might be different from those for supporting running of the
> binary).


The trouble is that despite carefully-written and lengthy documentation,
these requirements aren't fully documented, and yet it seems users are
expected to build Emacs for themselves, as the most recent release on the
DJGPP site is 23.3, and the emacs.README file on the download site says:

"Emacs can be compiled with DJGPP out of the box, and you are encouraged to
get the latest version from the GNU ftp sites and build it by yourself."

So either that expectation should be changed (i.e. binary builds should be
supplied in a timely fashion to the DJGPP site, and that sentence in the
documentation changed), or the build instructions should be improved.

We could try debugging those problems, but unless you intend to continue
> building
> the DJGPP port, that, too, would be a waste, IMO.
>

If people are still using Emacs on DOS, and they're expected to build it
from source, surely this is not a waste of effort? Or, no-one is using it,
and we should remove the port.


> Given the above, investing efforts in moving DJGPP to the Posix
> configury is a waste of resources that are best applied to parts of
> Emacs that will benefit users of modern systems.
>

FreeDOS is an active project, and it's free. It seems odd to me that we
should be happy that Emacs builds on a defunct proprietary system, and not
(apparently) on a closely-related, active, free one.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 2842 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Sat, 9 Aug 2014 10:11:54 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>, 18205@debbugs.gnu.org
> 
> > The current DJGPP build "just works", given that you have a correct
> > setup on a system that supports the build (requirements for supporting
> > the build might be different from those for supporting running of the
> > binary).
> 
> The trouble is that despite carefully-written and lengthy documentation,
> these requirements aren't fully documented

They are documented for systems that the people who wrote those
instructions had access to.  FreeDOS is not among them, so I don't
know if it complies.

> and yet it seems users are expected to build Emacs for themselves

Not "expected", "encouraged".

> So either that expectation should be changed (i.e. binary builds should be
> supplied in a timely fashion to the DJGPP site, and that sentence in the
> documentation changed), or the build instructions should be improved.

I supplied the 23.3 binary.  I intend to upload a 24.4 binary once it
is out.  Previous versions were IMO not stable enough to warrant an
upload.

As for improving the instructions, I don't know what to do about
that.  Once again, I have multiple success reports from people
building Emacs on various platforms that support DOS, and only one
failure report from you.  With all due respect, I don't think this is
a sign of inadequate instructions.

As I wrote earlier, if you are willing to continue building the DJGPP
port on FreeDOS on a more or less regular basis, I'm here to help you
debug whatever problems you originally had with your build.

> If people are still using Emacs on DOS, and they're expected to build it
> from source, surely this is not a waste of effort? Or, no-one is using it,
> and we should remove the port.

See above: there's a world between these 2 extremities.  Evidently,
people who have working DJGPP installations do succeed in building
Emacs, and there are still users who'd like to have newer binaries.

> > Given the above, investing efforts in moving DJGPP to the Posix
> > configury is a waste of resources that are best applied to parts of
> > Emacs that will benefit users of modern systems.
> >
> 
> FreeDOS is an active project, and it's free. It seems odd to me that we
> should be happy that Emacs builds on a defunct proprietary system, and not
> (apparently) on a closely-related, active, free one.

The DJGPP port of Emacs is currently supported by a single individual
-- yours truly.  I'm doing the minimum to keep the port alive, so that
I could invest most of my scarce free time to more broadly useful
Emacs tasks.  If you want to help in supporting the DJGPP port,
welcome aboard.  If not, I respectfully request that you trust me to
do the best I can in the little time I can afford devoting to this.
Currently, I don't plan to move my DJGPP development environment to
FreeDOS, I simply cannot afford such an effort (FreeDOS is
sufficiently different from MS-DOS and its other emulations to
necessitate a non-trivial effort).  I understand the theory behind the
urge to move to FreeDOS, but it will have to be done by someone else.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 1050 bytes --]

On 9 August 2014 12:53, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Sat, 9 Aug 2014 10:11:54 +0100
> > From: Reuben Thomas <rrt@sc3d.org>
> > Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>,
> 18205@debbugs.gnu.org
> >
> > > The current DJGPP build "just works", given that you have a correct
> > > setup on a system that supports the build (requirements for supporting
> > > the build might be different from those for supporting running of the
> > > binary).
> >
> > The trouble is that despite carefully-written and lengthy documentation,
> > these requirements aren't fully documented
>
> As for improving the instructions, I don't know what to do about
> that.  Once again, I have multiple success reports from people
> building Emacs on various platforms that support DOS, and only one
> failure report from you.  With all due respect, I don't think this is
> a sign of inadequate instructions.
>

Sure, I'll give it another go from scratch.

As for the rest, I'll see what practical help I can give.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1765 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Sat, 9 Aug 2014 12:57:02 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: Richard Stallman <rms@gnu.org>, Glenn Morris <rgm@gnu.org>, 18205@debbugs.gnu.org
> 
> As for the rest, I'll see what practical help I can give.

Thanks in advance.

bug#18205: Obsolete patterns in auto-mode-alist

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

MSDOG is our name for Microsoft's non-graphical operating system
that preceded Losedows.  It would be correct to replace "MSDOS"
with "MSDOG".

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call.

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Wed, 6 Aug 2014 00:58:04 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> 
>      ;; .emacs or .gnus or .viper following a directory delimiter in
>      ;; Unix, MSDOG or VMS syntax.
>      ("[]>:/\\]\\..*\\(emacs\\|gnus\\|viper\\)\\'" . emacs-lisp-mode)
>      ("\\`\\..*emacs\\'" . emacs-lisp-mode)
>      ;; _emacs following a directory delimiter
>      ;; in MsDos syntax
>      ("[:/]_emacs\\'" . emacs-lisp-mode)
> 
> There are two non-standard spellings:
> 
> s/MSDOG/MS-DOS/
> s/MsDos/MS-DOS/
> 
> Further, the second line looks like it will only match a .emacs in the root
> directory

No, it matches .emacs with no leading directories at all.

> Given that VMS support has been removed, can we replace the first two
> pattern lines with a line that uses the pattern used above:
> 
> \\(/\\|\\`\\)

No, that's not enough: it will miss "c:.emacs".

IMO, it's OK to remove the VMS-specific characters (AFAIK, the first 2
in the character class), but the rest should stay.

> And can we then also remove the third line, as MS-DOS syntax is not
> treated specially anywhere else in this default value?

How can you remove it, and still support _emacs as a file whose
normal-mode is emacs-lisp?  _emacs is supported on MS-DOS and
MS-Windows.  (I don't understand what you meant by "not supported
specially anywhere else".)

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 2032 bytes --]

On 6 August 2014 19:01, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Wed, 6 Aug 2014 00:58:04 +0100
> > From: Reuben Thomas <rrt@sc3d.org>
> >
> >      ;; .emacs or .gnus or .viper following a directory delimiter in
> >      ;; Unix, MSDOG or VMS syntax.
> >      ("[]>:/\\]\\..*\\(emacs\\|gnus\\|viper\\)\\'" . emacs-lisp-mode)
> >      ("\\`\\..*emacs\\'" . emacs-lisp-mode)
> >      ;; _emacs following a directory delimiter
> >      ;; in MsDos syntax
> >      ("[:/]_emacs\\'" . emacs-lisp-mode)
> >
> > There are two non-standard spellings:
> >
> > s/MSDOG/MS-DOS/
> > s/MsDos/MS-DOS/
> >
> > Further, the second line looks like it will only match a .emacs in the
> root
> > directory
>
> No, it matches .emacs with no leading directories at all.
>

I thought these patterns were matching absolute filenames, in which case I
am not sure what "no leading directories" could mean (at least on GNU).

> Given that VMS support has been removed, can we replace the first two
> > pattern lines with a line that uses the pattern used above:
> >
> > \\(/\\|\\`\\)
>
> No, that's not enough: it will miss "c:.emacs".
>
> IMO, it's OK to remove the VMS-specific characters (AFAIK, the first 2
> in the character class), but the rest should stay.
>

OK.


> > And can we then also remove the third line, as MS-DOS syntax is not
> > treated specially anywhere else in this default value?
>
> How can you remove it, and still support _emacs as a file whose
> normal-mode is emacs-lisp?  _emacs is supported on MS-DOS and
> MS-Windows.  (I don't understand what you meant by "not supported
> specially anywhere else".)
>

Sorry, I was unclear: I meant that there doesn't seem to be special support
for MS-DOS and Windows filenames elsewhere *in auto-mode-alist* (as I put
it above, "in this default value"). If "_emacs" is to be supported, surely
other similar patterns should be added for other dot files? (The same
applies to "c:.emacs"; but again, that's not an absolute filename, which
would be "c:\.emacs".)

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 3152 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Wed, 6 Aug 2014 23:05:33 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: 18205@debbugs.gnu.org
> 
> If "_emacs" is to be supported, surely other similar patterns should
> be added for other dot files?

Which patterns are those?

Why do you think we support _emacs for DOS and Windows?

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 1096 bytes --]

On 7 August 2014 03:51, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Wed, 6 Aug 2014 23:05:33 +0100
> > From: Reuben Thomas <rrt@sc3d.org>
> > Cc: 18205@debbugs.gnu.org
> >
> > If "_emacs" is to be supported, surely other similar patterns should
> > be added for other dot files?
>
> Which patterns are those?
>

For example, the patterns for bash's dot-files.


> Why do you think we support _emacs for DOS and Windows?
>

I'm not sure why you're asking me, you're the DJGPP maintainer! If you
meant it as an exercise, then I'd say: I have no idea where we support it
on Windows, where, post-Win 3.x (which we don't support), ".emacs" is a
valid filename.

I understood that DJGPP had long filename support too, but maybe that's not
universal? In which case, not only "_emacs", but other GNU dotfiles should
surely be supported with a leading "_"?

In other words, it looks like a quick hack to support the most
commonly-edited Emacs file that would suffer from this problem. I can't
tell whether it's still needed, and if it is, it should be extended
systematically.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1920 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Eli Zaretskii]

> Date: Thu, 7 Aug 2014 13:02:26 +0100
> From: Reuben Thomas <rrt@sc3d.org>
> Cc: 18205@debbugs.gnu.org
> 
> On 7 August 2014 03:51, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > > Date: Wed, 6 Aug 2014 23:05:33 +0100
> > > From: Reuben Thomas <rrt@sc3d.org>
> > > Cc: 18205@debbugs.gnu.org
> > >
> > > If "_emacs" is to be supported, surely other similar patterns should
> > > be added for other dot files?
> >
> > Which patterns are those?
> >
> 
> For example, the patterns for bash's dot-files.

It was never required, and nowadays the MS-DOS build does not get any
features added, especially if no one asked for them.

> > Why do you think we support _emacs for DOS and Windows?
> >
> 
> I'm not sure why you're asking me, you're the DJGPP maintainer! If you
> meant it as an exercise, then I'd say: I have no idea where we support it
> on Windows, where, post-Win 3.x (which we don't support), ".emacs" is a
> valid filename.

We support _emacs on Windows because the Windows Explorer used to
dislike files with a leading dot, pretending they didn't exist or
couldn't be created.

A release or two ago, we deprecated _emacs on Windows, but we still
didn't remove support for it.

> I understood that DJGPP had long filename support too, but maybe that's not
> universal?

Only on Windows, or with a special driver.  Not on plain DOS.

> In which case, not only "_emacs", but other GNU dotfiles should
> surely be supported with a leading "_"?

In principle, yes.  But this is all ad-hoc, not universal.

> In other words, it looks like a quick hack to support the most
> commonly-edited Emacs file that would suffer from this problem. I can't
> tell whether it's still needed, and if it is, it should be extended
> systematically.

See above: _emacs is still needed.

bug#18205: Obsolete patterns in auto-mode-alist

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 651 bytes --]

On 6 August 2014 23:05, Reuben Thomas <rrt@sc3d.org> wrote:

> On 6 August 2014 19:01, Eli Zaretskii <eliz@gnu.org> wrote:
>
>> > Date: Wed, 6 Aug 2014 00:58:04 +0100
>> > From: Reuben Thomas <rrt@sc3d.org>
>> >
>
> > Given that VMS support has been removed, can we replace the first two
>> > pattern lines with a line that uses the pattern used above:
>> >
>> > \\(/\\|\\`\\)
>>
>> No, that's not enough: it will miss "c:.emacs".
>>
>> IMO, it's OK to remove the VMS-specific characters (AFAIK, the first 2
>> in the character class), but the rest should stay.
>>
>
> OK.
>

Installed as part of r117663. Thanks for the fix.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1589 bytes --]

bug#18205: Obsolete patterns in auto-mode-alist

[Glenn Morris]

Reuben Thomas wrote:

> The default value for auto-mode-alist currently contains the following:
>
>      ;; .emacs or .gnus or .viper following a directory delimiter in
>      ;; Unix, MSDOG or VMS syntax.
>      ("[]>:/\\]\\..*\\(emacs\\|gnus\\|viper\\)\\'" . emacs-lisp-mode)
>      ("\\`\\..*emacs\\'" . emacs-lisp-mode)
>      ;; _emacs following a directory delimiter
>      ;; in MsDos syntax
>      ("[:/]_emacs\\'" . emacs-lisp-mode)
>
> There are two non-standard spellings:
>
> s/MSDOG/MS-DOS/

It's a hilarious joke that certainly remains as side-splitting and
relevant today as it was in the 80's.