patch applied (ghc): Set -Wall -fno-warn-name-shadowing in compiler/ when stage /= 2

[Simon Peyton-Jones]

| > Any reason why /= 1?  Why not always?
|
| To avoid failures caused by erroneous warnings in old GHCs. Although
| given most of the time validate will be using the most recent released
| GHC, maybe turning warnings on for stage 1 too would be worthwhile so
| that if it's going to fall over it does so quickly. I'm happy either
| way.

My sense is that it'd be good to have stage1 warnings on. They do no harm, after all.


| It's a pity there isn't a -Wwarn flag, so we can't easily negate -Werror
| in GhcStage1Opts (or whatever it's called) only. I should add one now so
| that in the future we won't have the same problem.

Indeed that'd be a good thing. Flags should be negatable.

S