Re: eb fixes. Philsophy on broken code?

[Martin von Loewis <martin at mira dot isdn dot cs dot tu-berlin dot de>]

> I don't feel strongly about this, but believe there is some value in
> testing broken source.  No, I wouldn't want to get carried away with 
> the idea.

Neither do I. You are certainly right: errors in code need to be
detected, and the testsuite must check whether they get detected.  In
some of your cases, you indicate that earlier versions plain crash, so
the test cases should stay.

In some cases, I felt that g++ should have detected the error for
ages, and the report was just bogus. I might be wrong.

Anyway, I think that you did an incredible job with these.

Thanks,
Martin