RE: Regression regarding Storable

I couldn't' resist looking into this bug this evening.  It turned out to be a consequence of a mistake in typechecking lazy patterns, which only showed up with the new feature I've just added (allowing constructors to bind dictionaries).

Thanks for finding it; it should work now.

Simon

| -----Original Message-----
| From: [email protected] [mailto:[email protected]] On Behalf Of Sven Panne
| Sent: 24 November 2006 17:39
| To: [email protected]
| Subject: Regression regarding Storable
|
| I've compiled GHC from HEAD this morning, and it once again broke my OpenGL
| binding. After boiling down the example (see attached Foo.hs), it seems that
| either Storable itself, constant folding or the strictness analysis is buggy:
|
|    panne@spock:~> ./Foo
|    *** main 1
|    *** main 2
|    Foo: Prelude.undefined
|
| I suspect that the "undefined" is the one in the default instance of
| peekElemOff, but that is just guessing. GHC even goes so far to optimize away
| the last putStrLn (see Foo.simpl). :-(
|
| Cheers,
|    S.