This seems to be the most common way that validate is accidentally
broken, these days -- not testing with both DEBUG defined and undefined.
Rather than double the testing time, I suggest validate is modified to
either:
- Compile one of stage1 and stage2 with DEBUG and the other without.
- But if you want to compile them both with DEBUG (or both without),
then, finish up by compiling a stage3 (or whichever "stage") with the
opposite sense of DEBUG, and without any optimizations because you're
not going to be using it.
These won't catch behavior differences that only testing can catch --
but I think most of these errors are just incorrect imports etc., so I'm
suggesting something that seems expedient to me. There are still
buildbots to help catch the rest after-the-fact.
what do you think?
-Isaac
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
