-fsquangle

[Jeffrey A Law]

  In message < 199811271658.IAA01115@adsl-206-170-148-33.dsl.pacbell.net >you wri
te:
  > >>>>> "Andrew" == Andrew Macleod <amacleod@cygnus.com> writes:
  > 
  >     Andrew> To the bets of my knowledge, it was just the 'n' counter.
  >     Andrew> Mark has done a lot of stuff to squangling since, and I
  >     Andrew> was under the impression he has corrected all this. True
  >     Andrew> Mark? 
  > 
  > I believe I have made all necessary changes on both the mangling and
  > unmangling sides to get demangling to work correctly.  This was, at
  > least, true for a while on some rather large C++ programs (millions of
  > lines); the demangled output was the same with and without -fsquangle.
  > 
  > Unfortunately, there's no easy way to write test-cases for this, given
  > our current testing framework, so we don't have a way to test that we
  > haven't broken stuff.  If someone contributed a new test-harness piece
  > that allowed one to compile the same source file with and without
  > -fsquangle, run `nm | c++filt' on the resulting .o files, and compare
  > the results that would be very nice.
An indirect way to test this would be to use the demangling tests in the gdb
testsuite.

jeff