small project idea (~30 lines)

[Mike Stump]

> Date: Mon, 08 Mar 1999 15:24:32 -0700
> From: Jeffrey A Law <law@hurl.cygnus.com>

> ie, I know with reasonable accuracy what tests fail on targets I
> work with a lot, and they're mostly restricted to tests added since
> mid 1998.  If something from 1997 pops up, then I'll make the
> assumption that I've broken something and that it needs to be fixed
> asap before I lose too much context and forget the details of
> whatever recent changes I've made.

If someone wants to help out, and find a way to mark the expected
failures in the testsuite that the maintainers expect, this I feel
would be a big help.  It had been a goal of mine since 1992 for g++,
and g++ is there (subject to a random target or a random 4 new tests
that start with the letters conpr-) already.  With gcc, I note that I
still get a canonical 36-59 failures.  The problem is, with any given
run, I cannot determine what to expect, so I loose (we loose) the fine
grain resolution the testsuite _should_ be able to provide.

The dejagnu infrastructure has all the required semantics in it, we
just need a syntax and a syntax processor.

This ability would bring gcc in line with g++, objc, fortran, g++,
libstdc++ and libio.

I'd be happy to point out the place where it goes and current code
that does what one needs to do and the design work necessary to come
up with a solution, if people need some help.  An ability to work in
TCL is required, which is the most horrible part of the job.  :-)