This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Proposed targets to obsolete for 3.4, first pass


This is my tentative list of targets to make obsolete for 3.4 (and
remove from 3.5), with notes.  This is the first pass; I am soliciting
comments and changes from maintainers.

This is mildly agressive in some ways, and I expect people to ask for the
retention of some of these.  It's rather conservative in other ways, so I
expect someone to add suggestions too.

I have not gone after most of the aout and coff targets.  I also have
not gone after most of the -bsd and -sysv targets, on the assumption 
that they are most likely used as templates for new targets rather than
as actual targets themselves.

If people volunteer that they are using or intend to use GCC 3 
on one of these systems, that system is more likely to be reprieved.

For the specific triples, if someone volunteers to run regular builds
and testsuite runs of GCC on one of these systems, and to keep in touch
with maintainers trying to debug problems, it will probably be reprieved.
(This is why i?86-sequent-ptx4 isn't on this list -- I have a guinea pig
who is doing the testing while I try to get it build.)

If someone volunteers that they will maintain a given
system, why then it will definitely be reprieved.  :-)

Note that the following targets have maintainers who have volunteered
but haven't made it into MAINTAINERS yet:
arc -- Rich D'Addio
frv -- Eric Christopher

--
Chips to obsolete all variants of:

d30v-*
  No maintainer, no documented instruction set

dsp16xx-*
  No maintainer.  Michael Collison (collison@isisinc.net) contributed it and
  claims to be the "owner" -- but isn't a gcc maintainer at the moment.
  The only changes since Oct 2001 have been updates for various systematic
  changes.  No test results ever reported to gcc-testresults.

  It also lacks nearly all the nice features on Zack's list of nice
  features to have.

  If someone can give a test report that this works in mainline, I'll take
  it off the list.

i370-*-*
  Yes, this supports older hardware than s390 does.  Yes, it supports
  EBCDIC.  Yes, it supports MVS.  Yes, it supports OpenEdition. 

  It doesn't have a maintainer.  It doesn't have any test results
  reported.  It hasn't had any serious work done since 1999, despite the
  various dire warnings about how it might not work.  Supporting EBCDIC
  for the host creates lots of work for the general maintainers.

i960-*
  Maintainer says it's broken and isn't going to get fixed.

--
Specific target triples to obsolete for hppa:
hppa1.0-*:  No test results.  Adds gratuituous complication.  Everyone seems
to use hppa1.1 machines or better.

hppa*-*-pro*:  No test results.  Obsoleted by GDB because "nobody is using" it.
hppa*-*-osf*:  No test results.  Obsoleted by GDB because "nobody is using" it.

--
Specific target triples to obsolete for i?86:

i?86-moss-msdos, i?86-*-moss*:
  The moss-msdos triple was added by Bryan Ford to GCC in 1996.  The other
  one appears to be ancient.  I can't find any evidence on the web
  that this system even exists any more.  And it causes me trouble in
  certain cleanups.  It's also apparently never produced by config.guess.

i?86-ncr-sysv4*:
  No test results ever reported.  Quite likely to be broken.  Very
  similar to generic i?86-sysv4.

i?86-*-netware:
  No test results ever reported.  If someone reports that this works,
  I'll take it off this list.

i?86-*-freebsd2*, i?86-*-freebsd*aout*
  Last test results in 2001.  If they're not using a newer freebsd....

i?86-*-linux*aout*:
  No test results in recent history.
  If they're not using ELF....

i?86-*-linux*libc1*:
  No test results in recent history.
  If they're not using glibc...

i?86-*-interix (not interix 3):
  No test results since gcc 2.95.  If they're not using Interix 3....

i?86-*-lynxos*:
  No test results ever reported.  Would allow the complete removal of
  quite a few files.

i?86-*-mach*:
  No test results ever reported.  Mach is gettin' kind of old and isn't
  a complete OS anyway....

i?86-*-udk*:
  Kean Johnston is taking care of this, but indicates that he would
  rather not be.  No test results since gcc "2.96". It's pretty unloved.

i?86-*-sysv[123]:
  No test results ever reported.  Unlikely to work at all.  Antique.

i386-*-vsta:
  No test results ever reported.  Had bugs which probably broke it for
  months with nobody noticing or complaining.

--
Specific target triples to obsolete for m68k:

These all have these features:  No test results ever reported.  Bernardo
Innocenti would like to get rid of it (and nobody complained).  See
http://gcc.gnu.org/ml/gcc/2003-09/msg00535.html.  Quite likely to be broken.

m68k-hp-hpux*, m68000-hp-hpux*
m68k*-*-netbsd* (NetBSD < 1.4 with a.out)
m68k-*-sysv4*

--
Specific target triples to obsolete for powerpc:

powerpc-*-chorusos*:
  No test results ever reported.  Doubtful that it works....

powerpc-wrs-windiss*:
  No evidence that this target ever existed, except for Mark Mitchell
  checking it in in 2002.

rs6000-*-lynxos*:
  No test results ever reported.

--
Specific target triples to obsolete for VAX:

vax-*-ultrix*:
  Last test results from gcc "2.97".
vax-*-*:
  (Default entry only)  Likely doesn't work, contains no useful information.

-- 
Nathanael Nerode  <neroden at gcc.gnu.org>
http://home.twcny.rr.com/nerode/neroden/fdl.html


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]