Aha. There was an inconsistency in the production of interfaces.
Without -O we suppress all 'pragma' info in interface files.
But we weren't suppressing rules, and one rule mentioned 'badHead'.
Because some other part thought we *were* suppressing the rules,
the defn of badHead didn't come out.
This'll be fixed when I do Simon's Big Commit
Workaround: compile the prelude with -O
Simon
| -----Original Message-----
| From: George Russell [mailto:[EMAIL PROTECTED]]
| Sent: 09 March 2000 14:42
| To: [EMAIL PROTECTED]
| Subject: PrelList.badHead - has GHC got a hangover?
|
|
| Sadly my shiny new GHC (downloaded and compiled from CVS last
| night) seems to
| have produced an inconsistent .hi file (attached) when
| compiling the basis.
| The problem is that badHead is mentioned in a inline bit of gunk, but
| the type for it isn't specified anywhere. (It's an internal
| value which
| doesn't get exported.)
|
