Drop OS X 10.4

Drop OS X 10.4

[Glenn Morris]

I suggest Emacs support for OS X 10.4 be explicitly dropped.
The OS itself is unsupported for 5 years.

There are 6 open Emacs bugs about it, all from a single person.
Half of them say it doesn't build, or crashes immediately.
All but one are unanswered, which suggests we already dropped it implicitly.

Re: Drop OS X 10.4

[Paul Eggert]

Glenn Morris wrote:
> I suggest Emacs support for OS X 10.4 be explicitly dropped.
> The OS itself is unsupported for 5 years.

Yes.  The general guideline in GNU packages is that when a platform's supplier 
no longer supports the platform, the GNU package doesn't need to support it 
either.  This is a reasonable guideline for Emacs too, e.g., our recent dropping 
support for IRIX 6.5 after SGI stopped support.  Since Apple has also dropped 
support for OS X 10.5 and 10.6, we should probably drop support for these 
versions as well.

Re: Drop OS X 10.4

[Glenn Morris]

Paul Eggert wrote:

> Yes.  The general guideline in GNU packages is that when a platform's
> supplier no longer supports the platform, the GNU package doesn't need
> to support it either.  This is a reasonable guideline for Emacs too,
> e.g., our recent dropping support for IRIX 6.5 after SGI stopped
> support.  Since Apple has also dropped support for OS X 10.5 and 10.6,
> we should probably drop support for these versions as well.

Hah, oh man, good luck trying to get Emacs to drop Windows 9x etc... ;)

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

10 okt 2014 kl. 17:59 skrev Glenn Morris <rgm@gnu.org>:

> 
> I suggest Emacs support for OS X 10.4 be explicitly dropped.
> The OS itself is unsupported for 5 years.
> 
> There are 6 open Emacs bugs about it, all from a single person.
> Half of them say it doesn't build, or crashes immediately.
> All but one are unanswered, which suggests we already dropped it implicitly.

FWIW, that was what I was aiming for when merging macfont.c.  So I totally agree
and planned to do this at some unknown future time (after the release probably).

I don't think anyone hacking on the ns-code has 10.4 available.

	Jan D.

Re: Drop OS X 10.4

[Paul Eggert]

Glenn Morris wrote:
> good luck trying to get Emacs to drop Windows 9x etc.

It's just a guideline, not an ironclad rule.  If a developer is amused by 
keeping Emacs working on an old platform and if this doesn't get in the way of 
the rest of us, it's fine.  But the default should be to drop junk that's 
present only for obsolete and unsupported platforms.  For example, there's no 
longer any need to clutter the code with support for K&R C.

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

10 okt 2014 kl. 18:18 skrev Paul Eggert <eggert@cs.ucla.edu>:

> Glenn Morris wrote:
>> I suggest Emacs support for OS X 10.4 be explicitly dropped.
>> The OS itself is unsupported for 5 years.
> 
> Yes.  The general guideline in GNU packages is that when a platform's supplier no longer supports the platform, the GNU package doesn't need to support it either.  This is a reasonable guideline for Emacs too, e.g., our recent dropping support for IRIX 6.5 after SGI stopped support.  Since Apple has also dropped support for OS X 10.5 and 10.6, we should probably drop support for these versions as well.

I would not want to drop 10.6, it still has significant usage, 18% in March according to:

http://www.macrumors.com/2014/03/27/os-x-mavericks-adoption-50/

Also, I have a 10.6 machine.  26.1 would perhaps be reasonable.

	Jan D.

Re: Drop OS X 10.4

[Harald Hanche-Olsen]

[Jan Djärv <jan.h.d@swipnet.se> (2014-10-10 17:13:09 UTC)]

> I would not want to drop 10.6, it still has significant usage, 18% in March according to:
> 
> http://www.macrumors.com/2014/03/27/os-x-mavericks-adoption-50/

Old PowerPC Macs cannot run later versions of the OS than 10.5. Perhaps that is why, according to the same page, about 3% of Macs are on that version.

I am not saying that support for 10.5 shouldn't be dropped too, just mentioning this as a possibly relevant fact.

– Harald

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

10 okt 2014 kl. 19:57 skrev Harald Hanche-Olsen <hanche@math.ntnu.no>:

> [Jan Djärv <jan.h.d@swipnet.se> (2014-10-10 17:13:09 UTC)]
> 
>> I would not want to drop 10.6, it still has significant usage, 18% in March according to:
>> 
>> http://www.macrumors.com/2014/03/27/os-x-mavericks-adoption-50/
> 
> Old PowerPC Macs cannot run later versions of the OS than 10.5. Perhaps that is why, according to the same page, about 3% of Macs are on that version.
> 
> I am not saying that support for 10.5 shouldn't be dropped too, just mentioning this as a possibly relevant fact.

We should drop PowerPC also.  It has been unsupported in practice for years.

	Jan D.

Re: Drop OS X 10.4

[Bozhidar Batsov]

[-- Attachment #1: Type: text/plain, Size: 845 bytes --]

Sounds reasonable to me. 

— 
Cheers,
Bozhidar

On October 11, 2014 at 10:24:48, Jan Djärv (jan.h.d@swipnet.se) wrote:

Hello.  

10 okt 2014 kl. 19:57 skrev Harald Hanche-Olsen <hanche@math.ntnu.no>:  

> [Jan Djärv <jan.h.d@swipnet.se> (2014-10-10 17:13:09 UTC)]  
>  
>> I would not want to drop 10.6, it still has significant usage, 18% in March according to:  
>>  
>> http://www.macrumors.com/2014/03/27/os-x-mavericks-adoption-50/  
>  
> Old PowerPC Macs cannot run later versions of the OS than 10.5. Perhaps that is why, according to the same page, about 3% of Macs are on that version.  
>  
> I am not saying that support for 10.5 shouldn't be dropped too, just mentioning this as a possibly relevant fact.  

We should drop PowerPC also. It has been unsupported in practice for years.  

Jan D.  



[-- Attachment #2: Type: text/html, Size: 1801 bytes --]

Re: Drop OS X 10.4

[Juergen Fenn]

2014-10-10 18:18 GMT+02:00 Paul Eggert <eggert@cs.ucla.edu>:

>  Since Apple
> has also dropped support for OS X 10.5 and 10.6, we should probably drop
> support for these versions as well.

Please note that there are still quite some 10.6.8 installations
around since it was the last version of Mac OS X to include Rosetta.

Regards,
Jürgen.

Re: Drop OS X 10.4

[Bozhidar Batsov]

[-- Attachment #1: Type: text/plain, Size: 719 bytes --]

And there’s the soon-to-be-released Emacs 24.4 that will work just fine on them. :-)

I doubt that people running such an old operating system have any illusions about being able to run all the latest and greatest software.

— 
Cheers,
Bozhidar

On October 11, 2014 at 15:59:32, Juergen Fenn (schneeschmelze@googlemail.com) wrote:

2014-10-10 18:18 GMT+02:00 Paul Eggert <eggert@cs.ucla.edu>:  

> Since Apple  
> has also dropped support for OS X 10.5 and 10.6, we should probably drop  
> support for these versions as well.  

Please note that there are still quite some 10.6.8 installations  
around since it was the last version of Mac OS X to include Rosetta.  

Regards,  
Jürgen.  


[-- Attachment #2: Type: text/html, Size: 1904 bytes --]

Re: Drop OS X 10.4

[Glenn Morris]

Jan, who does the work, and is therefore the only person whose opinion
really matters, has already said he will keep 10.6.

http://lists.gnu.org/archive/html/emacs-devel/2014-10/msg00336.html

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

Trunk now requires 10.6.  Older code is removed.  macfont.m is always used on COCOA,
nsfont.m only on GNUStep.  PowerPC unexec code is gone.

	Jan D.

11 okt 2014 kl. 20:25 skrev Glenn Morris <rgm@gnu.org>:

> 
> Jan, who does the work, and is therefore the only person whose opinion
> really matters, has already said he will keep 10.6.
> 
> http://lists.gnu.org/archive/html/emacs-devel/2014-10/msg00336.html

Re: Drop OS X 10.4

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 276 bytes --]

Jan Djärv wrote:

> Trunk now requires 10.6.

Thanks for doing all that.  I also squashed a few stray mentions of older 
versions in the documentation.  One little thing: the 'clear' method in nsmenu.m 
has a pre-10.6 hack; could it be simplified as per the attached?

[-- Attachment #2: macosx.patch --]
[-- Type: text/plain, Size: 544 bytes --]

=== modified file 'src/nsmenu.m'
--- src/nsmenu.m	2014-10-12 08:35:50 +0000
+++ src/nsmenu.m	2014-10-12 14:44:19 +0000
@@ -693,9 +693,7 @@
     {
       NSMenuItem *item = [self itemAtIndex: n];
       NSString *title = [item title];
-      if (([title length] == 0  /* OSX 10.5 */
-	   || [ns_app_name isEqualToString: title]  /* from 10.6 on */
-	   || [@"Apple" isEqualToString: title]) /* older */
+      if ([ns_app_name isEqualToString: title]
           && ![item isSeparatorItem])
         continue;
       [self removeItemAtIndex: n];

Re: Drop OS X 10.4

[Jan Djärv]

12 okt 2014 kl. 17:02 skrev Paul Eggert <eggert@cs.ucla.edu>:

> Jan Djärv wrote:
> 
>> Trunk now requires 10.6.
> 
> Thanks for doing all that.  I also squashed a few stray mentions of older versions in the documentation.  

Thanks.

> One little thing: the 'clear' method in nsmenu.m has a pre-10.6 hack; could it be simplified as per the attached?

Go ahead.  I will check on some OSX versions to be sure.  Just because the comment says 10.5 does not make it certain.  I don't recognize that particular hack.

	Jan D.

Re: Drop OS X 10.4

[Glenn Morris]

It seems like the OS X version is only tested for --with-ns builds.
Should it not also be tested for any build on OS X (X, no-X)?

Also this is the kind of thing to mention in NEWS.

Re: Drop OS X 10.4

[Glenn Morris]

Glenn Morris wrote:

> Also this is the kind of thing to mention in NEWS.

Sorry, it's already there. I was looking in the "non-free" section,
where I think this belongs.

Re: Drop OS X 10.4

[Jan Djärv]

13 okt 2014 kl. 01:10 skrev Glenn Morris <rgm@gnu.org>:

> Glenn Morris wrote:
> 
>> Also this is the kind of thing to mention in NEWS.
> 
> Sorry, it's already there. I was looking in the "non-free" section,
> where I think this belongs.

I can move it.  I puit it there because of similarity to the IRIX message.

	Jan D.

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

12 okt 2014 kl. 20:27 skrev Glenn Morris <rgm@gnu.org>:

> 
> It seems like the OS X version is only tested for --with-ns builds.
> Should it not also be tested for any build on OS X (X, no-X)?

I don't think so.  After all, if you are building for X or no toolkit, it is basically just UNIX with an unexec that works on Intel.  I don't think that we should prevent configurations that we in principle support (X, no-X, unexmacosx.c).  A check for PowerPC might be in order though as we don't support that.

I could add that, but I don't have any PowerPC to test it on.

	Jan D.

Re: Drop OS X 10.4

[Glenn Morris]

Jan Djärv wrote:

> I don't think so. After all, if you are building for X or no toolkit,
> it is basically just UNIX with an unexec that works on Intel. I don't
> think that we should prevent configurations that we in principle
> support (X, no-X, unexmacosx.c). A check for PowerPC might be in order
> though as we don't support that.
>
> I could add that, but I don't have any PowerPC to test it on.

Oh, but all the other points still apply:

OS X 10.4 itself is unsupported for 5+ years.
No Emacs developer has it to test on.
All the bug reports about it come from one person (who seems to be doing
an "explore the compilation parameter space" thing rather than actually
wanting to use it).
All the bug reports are about PPC;
  http://debbugs.gnu.org/cgi/pkgreport.cgi?include=subject%3A10.4;package=emacs
 (whether this means it is working fine on non-ppc, I have no data for! :) )

So personally I see no value to pretending it is still supported, and
think it might as well be explicitly dropped.

> A check for PowerPC might be in order though as we don't support that.

Should be easy?

*** configure.ac	2014-10-12 22:56:45 +0000
--- configure.ac	2014-10-13 16:47:29 +0000
***************
*** 557,563 ****
    *-apple-darwin* )
      case "${canonical}" in
        i[3456]86-* )  ;;
!       powerpc-* )    ;;
        x86_64-* )     ;;
        * )            unported=yes ;;
      esac
--- 557,563 ----
    *-apple-darwin* )
      case "${canonical}" in
        i[3456]86-* )  ;;
!       powerpc-* )    unported=yes ;;
        x86_64-* )     ;;
        * )            unported=yes ;;
      esac

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

13 okt 2014 kl. 18:48 skrev Glenn Morris <rgm@gnu.org>:

> Jan Djärv wrote:
> 
>> I don't think so. After all, if you are building for X or no toolkit,
>> it is basically just UNIX with an unexec that works on Intel. I don't
>> think that we should prevent configurations that we in principle
>> support (X, no-X, unexmacosx.c). A check for PowerPC might be in order
>> though as we don't support that.
>> 
>> I could add that, but I don't have any PowerPC to test it on.
> 
> Oh, but all the other points still apply:
> 
> OS X 10.4 itself is unsupported for 5+ years.
> No Emacs developer has it to test on.
> All the bug reports about it come from one person (who seems to be doing
> an "explore the compilation parameter space" thing rather than actually
> wanting to use it).
> All the bug reports are about PPC;
>  http://debbugs.gnu.org/cgi/pkgreport.cgi?include=subject%3A10.4;package=emacs
> (whether this means it is working fine on non-ppc, I have no data for! :) )
> 
> So personally I see no value to pretending it is still supported, and
> think it might as well be explicitly dropped.
> 

I have to do some other test, because the test relies on Cocoa dev environment being present.
Stand by.

>> A check for PowerPC might be in order though as we don't support that.
> 
> Should be easy?
> 
> *** configure.ac	2014-10-12 22:56:45 +0000
> --- configure.ac	2014-10-13 16:47:29 +0000
> ***************
> *** 557,563 ****
>    *-apple-darwin* )
>      case "${canonical}" in
>        i[3456]86-* )  ;;
> !       powerpc-* )    ;;
>        x86_64-* )     ;;
>        * )            unported=yes ;;
>      esac
> --- 557,563 ----
>    *-apple-darwin* )
>      case "${canonical}" in
>        i[3456]86-* )  ;;
> !       powerpc-* )    unported=yes ;;
>        x86_64-* )     ;;
>        * )            unported=yes ;;
>      esac

Thanks, I checked that in.

	Jan D.

Re: Drop OS X 10.4

[Jan Djärv]

Hello.

I have added a test to configure.  But I was wondering about crosscompiling.  We really do not support that, do we?

	Jan D.

14 okt 2014 kl. 07:08 skrev Jan Djärv <jan.h.d@swipnet.se>:

> Hello.
> 
> 13 okt 2014 kl. 18:48 skrev Glenn Morris <rgm@gnu.org>:
> 
>> Jan Djärv wrote:
>> 
>>> I don't think so. After all, if you are building for X or no toolkit,
>>> it is basically just UNIX with an unexec that works on Intel. I don't
>>> think that we should prevent configurations that we in principle
>>> support (X, no-X, unexmacosx.c). A check for PowerPC might be in order
>>> though as we don't support that.
>>> 
>>> I could add that, but I don't have any PowerPC to test it on.
>> 
>> Oh, but all the other points still apply:
>> 
>> OS X 10.4 itself is unsupported for 5+ years.
>> No Emacs developer has it to test on.
>> All the bug reports about it come from one person (who seems to be doing
>> an "explore the compilation parameter space" thing rather than actually
>> wanting to use it).
>> All the bug reports are about PPC;
>> http://debbugs.gnu.org/cgi/pkgreport.cgi?include=subject%3A10.4;package=emacs
>> (whether this means it is working fine on non-ppc, I have no data for! :) )
>> 
>> So personally I see no value to pretending it is still supported, and
>> think it might as well be explicitly dropped.
>> 
> 
> I have to do some other test, because the test relies on Cocoa dev environment being present.
> Stand by.
> 
>>> A check for PowerPC might be in order though as we don't support that.
>> 
>> Should be easy?
>> 
>> *** configure.ac	2014-10-12 22:56:45 +0000
>> --- configure.ac	2014-10-13 16:47:29 +0000
>> ***************
>> *** 557,563 ****
>>   *-apple-darwin* )
>>     case "${canonical}" in
>>       i[3456]86-* )  ;;
>> !       powerpc-* )    ;;
>>       x86_64-* )     ;;
>>       * )            unported=yes ;;
>>     esac
>> --- 557,563 ----
>>   *-apple-darwin* )
>>     case "${canonical}" in
>>       i[3456]86-* )  ;;
>> !       powerpc-* )    unported=yes ;;
>>       x86_64-* )     ;;
>>       * )            unported=yes ;;
>>     esac
> 
> Thanks, I checked that in.
> 
> 	Jan D.
> 

Re: Drop OS X 10.4

[Glenn Morris]

Jan Djärv wrote:

> I have added a test to configure.

IIUC, I don't think you need to call uname -r again, because that
information is already in $canonical. build-aux/config.guess does:

  UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown

and then

  ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}

So I think it can just be merged into the main $canonical "unported"
test (see eg handling of sunos versions).

> But I was wondering about crosscompiling. We really do not support
> that, do we?

IIRC, we don't support that at all, anywhere.

Re: Drop OS X 10.4

[Jan Djärv]

14 okt 2014 kl. 19:26 skrev Glenn Morris <rgm@gnu.org>:

> Jan Djärv wrote:
> 
>> I have added a test to configure.
> 
> IIUC, I don't think you need to call uname -r again, because that
> information is already in $canonical. build-aux/config.guess does:
> 
>  UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown
> 
> and then
> 
>  ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}
> 
> So I think it can just be merged into the main $canonical "unported"
> test (see eg handling of sunos versions).

Ok, done.

	Jan D.

Re: Drop OS X 10.4

[Ivan Kanis]

October, 10 at 11:59 Glenn wrote:

> I suggest Emacs support for OS X 10.4 be explicitly dropped.
> The OS itself is unsupported for 5 years.

Hi Glenn,

I noticed when it stopped working (a display problem) and filed a bug.
It's fine if you stop supporting 10.4. I spent way too many hours on
that Mac anyway. I have decided to shelve it.

Ivan

-- 
Emissions from cellular phones.
    -- BOFH excuse #8