DWARF-1 gcc status

[Michael Elizabeth Chastain <mec at shout dot net>]

I spent some time looking at gcc targets that preferred dwarf-1.
Technically the name is 'dwarf' but I prefer to say 'dwarf-1' to be
clear.

In gcc 3.2.1 the following targets prefer dwarf-1:

  i[34567]86-dg-dgux*		# obsolete in gcc 3.2.1, removed in gcc 3.3
  i[34567]86-sequent-ptx4*	# TD-R2
  i[34567]86-sequent-sysv4*	# TD-R2
  mips-sni-sysv4		# TD-R2
  m88k-dg-dgux*			# TD-R2
  sparc-hal-solaris2*		# TD-R2

There are many other targets which can generate dwarf-1 but prefer
to generate dwarf-2:

  many configurations with elfos.h
  mcore-*-elf
  sparclite-*-elf*
  sparc86x-*-elf*

Some targets can generate dwarf-1 but prefer to generate dbx:

  arc-*-elf*
  i[34567]86-*-osfrose*
  m32r-*-elf*
  mips-dec-osfrose*

Some targets can generate dwarf-1 but prefer to generate sdb:

  i[34567]86-*-scov3.2v5*
  m88k-dg-dgux* # TD-R2

The targets marked 'TD-R2' are in Zack Weinberg's list of
"Target Deprecation, Round 2".  This is a candidate list of targets
to be deprecated in gcc 3.3 and removed in gcc 3.4.

  http://gcc.gnu.org/ml/gcc/2002-12/msg00702.html

The gotcha is that gcc 2.95.3 supported a lot of targets with dwarf-1
and there may be a lot of people still use gcc 2.95.3 precisely because
gcc 3.X removed their favorite dwarf-1 target.  So we may have to support
dwarf-1 as long as we support gcc 2.95.3.

I filed a PR "maybe obsolete i[3456]86-dg-dgux*" but I don't have time
to do the field work of finding out if these systems are still in the
field.  EMC bought Data General in August 1999, and a year later,
a DG exec said that they would keep supporting Unix, but their main
focus is Windows NT.

Michael C