false regressions (was Re: results for egcs-2.91.53 586-pc-linux-gnulibc1)

[Mike Stump]

> It may be possible to get the information out in another way,

> I would propose to add
> FAIL: g++.robertl/eb125.C (test for internal compiler error)

I feel this is to marginal to be worth it.  I guess I don't feel
strongly about it, I do worry that a simple rule like `if it passed
before and now fails, we don't release' can't work with the above
solution, and I do like the simple rules for interpreting testcase
results.  :-)

That said, if you like the above solution over the mark them fail, I
won't object, though I do prefer the `mark them as fails' solution I
guess.