cvs commit: fptools/libraries/base/GHC Base.lhs

[Ross Paterson]

On Fri, Feb 13, 2004 at 03:26:21PM -0000, Simon Peyton-Jones wrote:
> GHC passes type arguments, and it must pass them in the right order,
> else all hell breaks loose.  At a call site GHC generates
> 
> 	(>>=)  t1 t2 foo (\x -> baz)
> 
> and t1, t2 had better be in the right order.
> 
> It's an internal-to-GHC thing really.

Right, but I'm also interested in GHC's desugaring.  The old deal was
that the argument order was the order of the leftmost occurrences of the
variables in the type.  Is that changing, or is this protection against
any change in the future?