Getting gdb and g++ to work together again

[Michael Elizabeth Chastain <chastain at cygnus dot com>]

The g++/*.exp tests are down to about 180 failures.  I'm getting past
the trivial "char*" and "(void)" issues into places where I actually want
changes in gcc.

I'm planning to manage these by:

. opening a gdb PR for each issue

  Even if there is no engineering work in gdb, there's a lot of state to
  track over time in getting the issues resolved.

. assigning the PR's to myself

  Anyone else can take the PR's if they want to join in.

. investigating down to the stabs level

  I want to present the issue as "I want a change in your stabs",
  because that's the boundary of what gcc is resposible for.

. talking to gcc people on gcc, gcc-bugs, and gcc-patches

. opening gcc PR's as needed

. pushing to get the gcc PR's fixed or resolved

. when gcc fixes a gcc PR, the person with the gdb PR can make any necessary
  gdb changes and gdb testsuite changes, test the new gcc+gdb combination,
  and then close the gdb PR.

. when gcc resolves a gcc PR by saying "we're going to keep emitting what we
  emit, deal with it", the person with the gdb PR can either (a) adapt gdb
  to match, or (b) push back harder to get gcc to change it on their side.

My first issue is stabs for methods marked "*INTERNAL*".  Jason Merrill
wrote a patch on 2001-02-10, but it's not in the gcc sources.

Michael