removing support for old platforms

removing support for old platforms

[Dan Nicolaescu]

RMS has OKed removing support for the platforms in the list below.
(after advertising this on a few relevant list and waiting a few
months).

In order to not leave junk in the tree, we could collect all the symbols
it defines in a new section at the one of admin/CPP-DEFINES. If all the
#defines for a symbol have been removed, then the code that makes
use of that define can be removed.

The MSDOS platform still has some interesting code that we might want to
reuse: it has real menus in text mode, that functionality would be good
to have at some point in the future for terminals.
Eli, how do you wan to handle this?

Help doing this work would be greatly appreciated.

MSDOS
Mac OS all versions before MACOSX (only MAC_OS8 seems to have a special define)

Apollo SR10.x (unexapollo.c)
Convex (unexconvex.c and m/convex.c)
Xenix (unexenix.c and s/xenix.h)
Iris  (unexmips.c m/iris4d.h   m/irist.h    s/iris3-5.h  s/iris3-6.h)
Gould (m/gould*)
Siemens machines running Sinix (unexsni.c)
Harris CXUX (s/cxux*)
ESIX, a variant of v.5.3 for the 386. (s/esix*)
Interactive (ISC) Unix (s/isc*)
Sony News (s/newsos*)
RTU 3.0, ucb universe (s/rtu.h) -- defines NOMULTIPLEJOBS
UniSoft's UniPlus 5.2 (s/uniplus.h)
UMAX (s/umax.h)
AT&T UNIX PC model 7300 (m/7300.h)
Acorn
Alliant (m/alliant*)
Amdahl (m/amdahl*)
Altos 3068 Unix System V Release 2 (m/altos.h)
Apollo (m/apollo.h)
AT&T 3b (m/att3b.h)
Aviion (m/aviion*)
Celerity (m/celerity.h)
clipper (m/clipper.h)
convergent S series (m/cnvrgnt.h)
cydra (m/cydra5.h)
Motorola System V/88 machines (m/delta88k.h) -- gcc dropped support
                              for 88k a while ago
Bull DPX/2 range (m/dpx2.h)
Dual machines using unisoft port (m/dual.h)
Elxsi machine (running enix) (m/elxsi.h)
Fujitsu F301 machine (m/f301.h)
i860 (m/i860.h) gcc dropped support a while ago
ibm ps/2 aix386 (m/ibmps2-aix.h)
ISI 68000's (m/is*)
Masscomp 5000 series running RTU, ucb universe (m/masscomp.h)
Megatest 68000's (m/mega68.h)
Whitechapel Computer Works MG1 (ns16000 based) (m/mg1.h)
Harris Night Hawk Series 1200 and Series 3000 (m/nh3000.h m/nh4000.h)
ns16000 (m/ns16000.h)
National Semiconductor 32000, running Genix (m/ns32000.h)
TI Nu machines using system V (m/nu.h)
HLH Orion (m/orion.h m/orion105.h)
Paragon i860 (m/paragon.h)
PFU A-series (m/pfa50.h)
Plexus running System V.2 (m/plexus.h)
pyramid. (m/pyramid.h)
sh3el (m/sh3el.h)
Bull SPS-7 (m/sps7.h)
m/sr2k.h
Stride (m/stride.h)
Sun 1 (m/sun1.h)
Sun 2 (m/sun2.h)
SEQUENT SYMMETRY  m/symmetry.h
Tadpole 68k machines m/tad68k.h
tahoe m/tahoe.h
targon31 m/targon31.h
Tektronix* (m/tek4300.h m/tekxd88.h)
NCR Tower 32 running System V.2 (m/tower32.h)
NCR Tower 32 running System V.3 (m/tower32v3.h)
U-station (Nihon Unisys, SS5E; Sumitomo Denkoh, U-Station E30). (m/ustation.h)
Wicat (m/wicat.h)
Honeywell XPS100 running UNIX System V.2 (m/xps100.h)


Data General's DG/UX  (s/dgux*)
Irix before version 6
osf1 (s/osf*)
SunOS4 (s/sunos*)
RISCiX (s/riscix*)
SCO 3.2v4 (s/sco4.h)
SCO 3.2v5 (s/sco5.h)
Sun's 386-based RoadRunner (m/sun386.h)
Sun3 machines (m/sun3*)
 Integrated Solutions 386 machine (m/is386.h)

Re: removing support for old platforms

[Eli Zaretskii]

> From: Dan Nicolaescu <dann@ics.uci.edu>
> Date: Sat, 05 Jan 2008 00:43:35 -0800
> 
> The MSDOS platform still has some interesting code that we might want to
> reuse: it has real menus in text mode, that functionality would be good
> to have at some point in the future for terminals.
> Eli, how do you wan to handle this?

I'd prefer that the support for the MSDOS port not be removed until I
decide to stop working on it altogether.  After all, apart of the
msdos/ subdirectory, which can hardly be a nuisance to anyone, the
MSDOS support boils down to:

  . 5 DOS-specific C files
  . several __MSDOS__/__DJGPP__ cpp directives scattered through the
    common sources.

That doesn't seem too much.

Re: removing support for old platforms

[Dan Nicolaescu]

Dan Nicolaescu <dann@ics.uci.edu> writes:

  > RMS has OKed removing support for the platforms in the list below.
  > (after advertising this on a few relevant list and waiting a few
  > months).
  > 
  > In order to not leave junk in the tree, we could collect all the symbols
  > it defines in a new section at the one of admin/CPP-DEFINES. If all the
  > #defines for a symbol have been removed, then the code that makes
  > use of that define can be removed.

This is done now, except for
 - removing support for MSDOS  - Eli still wants it
 - removing support for Mac OS version before MACOSX - I don't know what
 exactly needs to be removed. Probably all the .mpw files, what else?
 And all the code that depends on MAC_OS8 being defined. 
 Help doing this would be appreciated. 

admin/CPP-DEFINES has a new section: 
** All references to these symbols should be removed, all their definitions are gone **
There are many files that refer to those symbols. Removing those
references would help.

There's also another section:
** Temporarily put defines from removed files here, check if these
   macros need to be documented or removed because they are not used
   anymore.  **  
Some of these might not be defined anywhere, then, as above, all the
references to them can be removed.

  > 
  > MSDOS
  > Mac OS all versions before MACOSX (only MAC_OS8 seems to have a special define)
  > 
  > Apollo SR10.x (unexapollo.c)
  > Convex (unexconvex.c and m/convex.c)
  > Xenix (unexenix.c and s/xenix.h)
  > Iris  (unexmips.c m/iris4d.h   m/irist.h    s/iris3-5.h  s/iris3-6.h)
  > Gould (m/gould*)
  > Siemens machines running Sinix (unexsni.c)
  > Harris CXUX (s/cxux*)
  > ESIX, a variant of v.5.3 for the 386. (s/esix*)
  > Interactive (ISC) Unix (s/isc*)
  > Sony News (s/newsos*)
  > RTU 3.0, ucb universe (s/rtu.h) -- defines NOMULTIPLEJOBS
  > UniSoft's UniPlus 5.2 (s/uniplus.h)
  > UMAX (s/umax.h)
  > AT&T UNIX PC model 7300 (m/7300.h)
  > Acorn
  > Alliant (m/alliant*)
  > Amdahl (m/amdahl*)
  > Altos 3068 Unix System V Release 2 (m/altos.h)
  > Apollo (m/apollo.h)
  > AT&T 3b (m/att3b.h)
  > Aviion (m/aviion*)
  > Celerity (m/celerity.h)
  > clipper (m/clipper.h)
  > convergent S series (m/cnvrgnt.h)
  > cydra (m/cydra5.h)
  > Motorola System V/88 machines (m/delta88k.h) -- gcc dropped support
  >                               for 88k a while ago
  > Bull DPX/2 range (m/dpx2.h)
  > Dual machines using unisoft port (m/dual.h)
  > Elxsi machine (running enix) (m/elxsi.h)
  > Fujitsu F301 machine (m/f301.h)
  > i860 (m/i860.h) gcc dropped support a while ago
  > ibm ps/2 aix386 (m/ibmps2-aix.h)
  > ISI 68000's (m/is*)
  > Masscomp 5000 series running RTU, ucb universe (m/masscomp.h)
  > Megatest 68000's (m/mega68.h)
  > Whitechapel Computer Works MG1 (ns16000 based) (m/mg1.h)
  > Harris Night Hawk Series 1200 and Series 3000 (m/nh3000.h m/nh4000.h)
  > ns16000 (m/ns16000.h)
  > National Semiconductor 32000, running Genix (m/ns32000.h)
  > TI Nu machines using system V (m/nu.h)
  > HLH Orion (m/orion.h m/orion105.h)
  > Paragon i860 (m/paragon.h)
  > PFU A-series (m/pfa50.h)
  > Plexus running System V.2 (m/plexus.h)
  > pyramid. (m/pyramid.h)
  > sh3el (m/sh3el.h)
  > Bull SPS-7 (m/sps7.h)
  > m/sr2k.h
  > Stride (m/stride.h)
  > Sun 1 (m/sun1.h)
  > Sun 2 (m/sun2.h)
  > SEQUENT SYMMETRY  m/symmetry.h
  > Tadpole 68k machines m/tad68k.h
  > tahoe m/tahoe.h
  > targon31 m/targon31.h
  > Tektronix* (m/tek4300.h m/tekxd88.h)
  > NCR Tower 32 running System V.2 (m/tower32.h)
  > NCR Tower 32 running System V.3 (m/tower32v3.h)
  > U-station (Nihon Unisys, SS5E; Sumitomo Denkoh, U-Station E30). (m/ustation.h)
  > Wicat (m/wicat.h)
  > Honeywell XPS100 running UNIX System V.2 (m/xps100.h)
  > 
  > 
  > Data General's DG/UX  (s/dgux*)
  > Irix before version 6
  > osf1 (s/osf*)
  > SunOS4 (s/sunos*)
  > RISCiX (s/riscix*)
  > SCO 3.2v4 (s/sco4.h)
  > SCO 3.2v5 (s/sco5.h)
  > Sun's 386-based RoadRunner (m/sun386.h)
  > Sun3 machines (m/sun3*)
  >  Integrated Solutions 386 machine (m/is386.h)

Re: removing support for old platforms

[Eli Zaretskii]

> From: Dan Nicolaescu <dann@ics.uci.edu>
> Date: Sun, 06 Jan 2008 09:46:22 -0800
> 
> Dan Nicolaescu <dann@ics.uci.edu> writes:
> 
>   > RMS has OKed removing support for the platforms in the list below.
>   > (after advertising this on a few relevant list and waiting a few
>   > months).
>   > 
>   > In order to not leave junk in the tree, we could collect all the symbols
>   > it defines in a new section at the one of admin/CPP-DEFINES. If all the
>   > #defines for a symbol have been removed, then the code that makes
>   > use of that define can be removed.
> 
> This is done now

Thanks!

Should we perhaps announce in NEWS the fact that these old
configurations are no longer supported?

Re: removing support for old platforms

[Dan Nicolaescu]

Eli Zaretskii <eliz@gnu.org> writes:

  > Should we perhaps announce in NEWS the fact that these old
  > configurations are no longer supported?

IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
it if you want to...

Re: removing support for old platforms

[Kim F. Storm]

Dan Nicolaescu <dann@ics.uci.edu> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>   > Should we perhaps announce in NEWS the fact that these old
>   > configurations are no longer supported?
>
> IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
> it if you want to...

Maybe it doesn't need to be specific, ie. it could just say

* Many pre-historic machines and systems are no longer supported.

(you may choose another term than pre-historic of course).

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: removing support for old platforms

[Richard Stallman]

Thanks for doing all that work.

Re: removing support for old platforms

[Lennart Borgman (gmail)]

Stephen J. Turnbull wrote:
> Kim F. Storm writes:
> 
>  > Dan Nicolaescu <dann@ics.uci.edu> writes:
>  > 
>  > > Eli Zaretskii <eliz@gnu.org> writes:
>  > >
>  > >   > Should we perhaps announce in NEWS the fact that these old
>  > >   > configurations are no longer supported?
>  > >
>  > > IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
>  > > it if you want to...
>  > 
>  > Maybe it doesn't need to be specific, ie. it could just say
>  > 
>  > * Many pre-historic machines and systems are no longer supported.
>  > 
>  > (you may choose another term than pre-historic of course).
> 
> More important, the list should be preserved somewhere, and a pointer
> to that list inserted here.


How about info antinews?

Re: removing support for old platforms

[Stephen J. Turnbull]

Kim F. Storm writes:

 > Dan Nicolaescu <dann@ics.uci.edu> writes:
 > 
 > > Eli Zaretskii <eliz@gnu.org> writes:
 > >
 > >   > Should we perhaps announce in NEWS the fact that these old
 > >   > configurations are no longer supported?
 > >
 > > IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
 > > it if you want to...
 > 
 > Maybe it doesn't need to be specific, ie. it could just say
 > 
 > * Many pre-historic machines and systems are no longer supported.
 > 
 > (you may choose another term than pre-historic of course).

More important, the list should be preserved somewhere, and a pointer
to that list inserted here.

Re: removing support for old platforms

[Eli Zaretskii]

> Date: Mon, 07 Jan 2008 22:59:29 +0100
> From: "Lennart Borgman (gmail)" <lennart.borgman@gmail.com>
> CC: "Kim F. Storm" <storm@cua.dk>, Eli Zaretskii <eliz@gnu.org>, 
>  Dan Nicolaescu <dann@ics.uci.edu>,
>  emacs-devel@gnu.org
> 
> Stephen J. Turnbull wrote:
> > Kim F. Storm writes:
> > 
> >  > Dan Nicolaescu <dann@ics.uci.edu> writes:
> >  > 
> >  > > Eli Zaretskii <eliz@gnu.org> writes:
> >  > >
> >  > >   > Should we perhaps announce in NEWS the fact that these old
> >  > >   > configurations are no longer supported?
> >  > >
> >  > > IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
> >  > > it if you want to...
> >  > 
> >  > Maybe it doesn't need to be specific, ie. it could just say
> >  > 
> >  > * Many pre-historic machines and systems are no longer supported.
> >  > 
> >  > (you may choose another term than pre-historic of course).
> > 
> > More important, the list should be preserved somewhere, and a pointer
> > to that list inserted here.
> 
> 
> How about info antinews?

No, not that one: Antinews is supposed to be an entertainment, no one
will ever seriously look there for any important information.

Re: removing support for old platforms

[Kim F. Storm]

Eli Zaretskii <eliz@gnu.org> writes:

>> How about info antinews?
>
> No, not that one: Antinews is supposed to be an entertainment, no one
> will ever seriously look there for any important information.

Well, it would be a good joke to read something like

* Emacs is now prepared to run on the following machines and systems,
which are expected to be available in some distant future: ...


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: removing support for old platforms

[David Kastrup]

storm@cua.dk (Kim F. Storm) writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> How about info antinews?
>>
>> No, not that one: Antinews is supposed to be an entertainment, no one
>> will ever seriously look there for any important information.
>
> Well, it would be a good joke to read something like
>
> * Emacs is now prepared to run on the following machines and systems,
> which are expected to be available in some distant future: ...

"in some distant past", please.

-- 
David Kastrup

Re: removing support for old platforms

[Kim F. Storm]

David Kastrup <dak@gnu.org> writes:

>> * Emacs is now prepared to run on the following machines and systems,
>> which are expected to be available in some distant future: ...
>
> "in some distant past", please.

If you live backwards in time, future seems like the right word...

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: removing support for old platforms

[Eli Zaretskii]

> Cc: "Lennart Borgman \(gmail\)" <lennart.borgman@gmail.com>,  stephen@xemacs.org,
> 	  dann@ics.uci.edu,  emacs-devel@gnu.org
> From: storm@cua.dk (Kim F. Storm)
> Date: Tue, 08 Jan 2008 10:12:59 +0100
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> How about info antinews?
> >
> > No, not that one: Antinews is supposed to be an entertainment, no one
> > will ever seriously look there for any important information.
> 
> Well, it would be a good joke to read something like
> 
> * Emacs is now prepared to run on the following machines and systems,
> which are expected to be available in some distant future: ...

Yes, but adding a long list of the configurations we dropped after
that will make it not very funny.

Re: removing support for old platforms

[Richard Stallman]

    Well, it would be a good joke to read something like

    * Emacs is now prepared to run on the following machines and systems,
    which are expected to be available in some distant future: ...

Yes that would be funny.  But I think "distant past" is what fits the
convention of Antinews better.  The convention is not that we call the
past the future.  Rather, we call it the past, but we talk about it in
the ways normally used for talking about the future.

The full list of discontinued platforms is not necessary for the joke
and would just be dull.  Maybe we could mention two or three.
Or maybe the joke doesn't need any specific names:

    * Emacs 21 supports many additional systems and computers
    that will be available in the distant past.

Re: removing support for old platforms

[Richard Stallman]

    No, not that one: Antinews is supposed to be an entertainment, no one
    will ever seriously look there for any important information.

Antinews has a serious purpose: to inform people who use the manual
with a previous version of the important differences from what the
manual says.  By "important" I mean differences they need to know
about.  The absence of new features, they will simply discover if they
try to use them, so these need not be described individually in
Antinews.

Re: removing support for old platforms

[Ted Zlatanov]

On Wed, 09 Jan 2008 06:37:42 -0500 Richard Stallman <rms@gnu.org> wrote: 

RS>     Well, it would be a good joke to read something like
RS>     * Emacs is now prepared to run on the following machines and systems,
RS>     which are expected to be available in some distant future: ...

RS> Yes that would be funny.  But I think "distant past" is what fits the
RS> convention of Antinews better.  The convention is not that we call the
RS> past the future.  Rather, we call it the past, but we talk about it in
RS> the ways normally used for talking about the future.

RS> The full list of discontinued platforms is not necessary for the joke
RS> and would just be dull.  Maybe we could mention two or three.
RS> Or maybe the joke doesn't need any specific names:

RS>     * Emacs 21 supports many additional systems and computers
RS>     that will be available in the distant past.

Here's my suggestion, avoiding "past" or "future" and keeping the
into-the-past idea of anti.texi:

@item
Great news!  Emacs 21 now supports an amazing number of systems and
computers soon to be available: [condensed list of systems and
computers]

I think having the full list is funnier, but it must be condensed so the
reader's eyes will glaze over and they will skip to the next item.
Otherwise it's annoying, if you can't easily skip forward when you're
tired of the joke (as I'm sure most of us are :).

Ted

Re: removing support for old platforms

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wed, Jan 09, 2008 at 07:48:38AM -0600, Ted Zlatanov wrote:
> On Wed, 09 Jan 2008 06:37:42 -0500 Richard Stallman <rms@gnu.org> wrote: 

[...]

> RS> The full list of discontinued platforms is not necessary for the joke
> RS> and would just be dull.  Maybe we could mention two or three.
> RS> Or maybe the joke doesn't need any specific names:

[...]

> I think having the full list is funnier, but it must be condensed so the
> reader's eyes will glaze over and they will skip to the next item.
> Otherwise it's annoying, if you can't easily skip forward when you're
> tired of the joke (as I'm sure most of us are :).

+1 on keeping (at least most of) the list. The names will evoke fond
memories (or is it premonitions -- I'm so confused :) on us oldies...

Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFHhfDcBcgs9XrR2kYRAnJoAJ9B7wuMh0muUtg8WyArK5ncZNbUQACfS6Wi
2mfeHL7hPs97WCb6Ehftcx4=
=c0mH
-----END PGP SIGNATURE-----

Re: removing support for old platforms

[Eli Zaretskii]

> From: Dan Nicolaescu <dann@ics.uci.edu>
> Cc: emacs-devel@gnu.org
> Date: Sun, 06 Jan 2008 12:44:15 -0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
>   > Should we perhaps announce in NEWS the fact that these old
>   > configurations are no longer supported?
> 
> IMHO, that would just add unnecessary bloat to NEWS. But feel free to do
> it if you want to...

Done (without bloat).

Re: removing support for old platforms

[Emanuele Giaquinta]

On Sun, Jan 06, 2008 at 09:46:22AM -0800, Dan Nicolaescu wrote:

> admin/CPP-DEFINES has a new section: 
> ** All references to these symbols should be removed, all their definitions are gone **
> There are many files that refer to those symbols. Removing those
> references would help.
> 
> There's also another section:
> ** Temporarily put defines from removed files here, check if these
>    macros need to be documented or removed because they are not used
>    anymore.  **  
> Some of these might not be defined anywhere, then, as above, all the
> references to them can be removed.

WORDS_BIG_ENDIAN is needed in md5.c and a couple of other files on big
endian systems.

Emanuele Giaquinta