Re: SunOS ports (M68k and sparc)

Re: SunOS ports (M68k and sparc)

[Robert Dewar]

It always seems to me that if people are happy to use obsolete unsupported
machines with obsolete unsupported operating systems, then it is reasonable
to expect that they use obsolete unsupported versions of GCC to go with it :-)

Re: SunOS ports (M68k and sparc)

[Joe Buck]

Zack Weinberg <zack@codesourcery.com> wrote:
> > I do not think we should restore SunOS -- or any other obsoleted
> > target -- just because someone offers to maintain the port...
> > someone maintain the port doesn't magically free all the rest of us
> > from the burden of keeping that port up to date with changes made to
> > target-independent code.  I think we should only restore such targets
> > if compelling reasons are presented why the entire GCC community
> > should bear the additional burden of supporting these old systems.

On Fri, Oct 17, 2003 at 07:42:17PM +0200, Jones Desougi wrote:
> Good points.
> With just me interested in these ports there is hardly compelling
> reasons.
> 
> As good as an older version of GCC may be, it will not support newer
> standards (or additional languages). It is still far better than the
> system compiler of course.

If the reason for your interest is to make use of some old hardware that
you have, you might want to investigate running Debian or BSD instead of
SunOS.

Re: SunOS ports (M68k and sparc)

[Jones Desougi]

Zack Weinberg <zack@codesourcery.com> wrote:
> ...
> 
> I do not think we should restore SunOS -- or any other obsoleted
> target -- just because someone offers to maintain the port.  Having
> someone maintain the port doesn't magically free all the rest of us
> from the burden of keeping that port up to date with changes made to
> target-independent code.  I think we should only restore such targets
> if compelling reasons are presented why the entire GCC community
> should bear the additional burden of supporting these old systems.

Good points.
With just me interested in these ports there is hardly compelling
reasons.

As good as an older version of GCC may be, it will not support newer
standards (or additional languages). It is still far better than the
system compiler of course.

   /Jones

Re: SunOS ports (M68k and sparc)

[Andi Kleen]

dewar@gnat.com (Robert Dewar) writes:

> It always seems to me that if people are happy to use obsolete unsupported
> machines with obsolete unsupported operating systems, then it is reasonable
> to expect that they use obsolete unsupported versions of GCC to go with it :-)


... especially since the many compile speed regressions will
make a modern gcc unusable on such a slow box anyways.

-Andi

Re: SunOS ports (M68k and sparc)

[Zack Weinberg]

Jones Desougi <jones@ingate.com> writes:

> On Wed, 15 Oct 2003 18:04:55 -0700, Mike Stump wrote:
>> Feel free to do up your paper work, and submit the changes to make it 
>> work.  We probably won't be interested in using out-dated C compilers, 
>> so, you will need a gcc binary to start with.
>> 
>> After you do the paper work, and contribute the work...  they just 
>> work, right?
>
> This gets a whole lot more messy if the existing pieces are removed in
> the meantime. That is why I asked it be taken off the obsoleted list.

I am afraid it's already gone from the mainline.

The targets we have been obsoleting are very old, and by virtue of
their age they cause significant extra work for GCC maintainers.
SunOS 4 is one of the targets whose runtime library predates the 1990
C standard.  These targets require us to maintain an extensive body of
code just to work around the bugs in their header files.  SunOS also
has the dubious distinction of having special case code in GCC to work
around bugs in its kernel -- this would not be an issue if you wanted
to support it solely as a cross-compile target, of course.

I do not think we should restore SunOS -- or any other obsoleted
target -- just because someone offers to maintain the port.  Having
someone maintain the port doesn't magically free all the rest of us
from the burden of keeping that port up to date with changes made to
target-independent code.  I think we should only restore such targets
if compelling reasons are presented why the entire GCC community
should bear the additional burden of supporting these old systems.

I am not absolutely ruling out the possibility of restoring SunOS 4
(I don't have the authority even if I wanted to).  However, Jones, I
would like to hear your reasons for wanting this, and especially your
reasons why you cannot continue to use older versions of GCC (which
may well work better than the current version, since none of us have
been designing with it in mind for years).

zw

Re: SunOS ports (M68k and sparc)

[Jones Desougi]

On Wed, 15 Oct 2003 18:04:55 -0700, Mike Stump wrote:
> Feel free to do up your paper work, and submit the changes to make it 
> work.  We probably won't be interested in using out-dated C compilers, 
> so, you will need a gcc binary to start with.
> 
> After you do the paper work, and contribute the work...  they just 
> work, right?

This gets a whole lot more messy if the existing pieces are removed in
the meantime. That is why I asked it be taken off the obsoleted list.

Re: SunOS ports (M68k and sparc)

[Mike Stump]

On Wednesday, October 15, 2003, at 04:20 PM, Jones Desougi wrote:
>> On Wed, Oct 15, 2003 at 02:31:42AM +0200, Jones Desougi wrote:
>>> I would like the sun systems with SunOS version 4 (m68k and sparc),
>>> also known as sun[34]-sunos4*, taken off the obsolete systems list.
>>
>> For that to happen, we would need a maintainer who would fix those 
>> ports
>> and keep them running as the compiler evolves.  Without that, saying
>> "OK, they are off the list" is meaningless, as they will still be 
>> broken.
>
> Quite so. There was a note of volunteering intended.

Feel free to do up your paper work, and submit the changes to make it 
work.  We probably won't be interested in using out-dated C compilers, 
so, you will need a gcc binary to start with.

After you do the paper work, and contribute the work...  they just 
work, right?

Re: SunOS ports (M68k and sparc)

[Jones Desougi]

> On Wed, Oct 15, 2003 at 02:31:42AM +0200, Jones Desougi wrote:
>> I would like the sun systems with SunOS version 4 (m68k and sparc),
>> also known as sun[34]-sunos4*, taken off the obsolete systems list.
> 
> For that to happen, we would need a maintainer who would fix those ports
> and keep them running as the compiler evolves.  Without that, saying
> "OK, they are off the list" is meaningless, as they will still be broken.

Quite so. There was a note of volunteering intended.

Re: SunOS ports (M68k and sparc)

[Joe Buck]

On Wed, Oct 15, 2003 at 02:31:42AM +0200, Jones Desougi wrote:
> I would like the sun systems with SunOS version 4 (m68k and sparc),
> also known as sun[34]-sunos4*, taken off the obsolete systems list.

For that to happen, we would need a maintainer who would fix those ports
and keep them running as the compiler evolves.  Without that, saying
"OK, they are off the list" is meaningless, as they will still be broken.

SunOS ports (M68k and sparc)

[Jones Desougi]

I would like the sun systems with SunOS version 4 (m68k and sparc),
also known as sun[34]-sunos4*, taken off the obsolete systems list.

     /Jones