Inferred type is less polymorphic than expected?

[Thomas Jäger]

Hi,

On 7/30/05, Benjamin Franksen <benjamin.franksen at bessy.de> wrote:
> On Saturday 30 July 2005 03:54, Thomas Jäger wrote:
> > You're probably refering to the version where you omit the type
> > signatures of both 'printer' and 'printCatalog'. Here the situation
> > is different because the compiler needs to infer the types and not
> > just check. In the explicitely typed version, 'printCatalog' is used
> > polymorphically, i.e. the 'c' is instantiated to different types
> > during the recursive call. Thus the example needs polymorphic
> > recursion, for which type inference is known to be undecidable. It is
> > therefore reasonable that ghc (as well as hugs) can't compile the
> > code (in fact, I guess type inference will assume that both contexts
> > are equal).
> 
> Hmm. I am not sure I understand this.
I just described the reason why the "inferred type..." error message
appears when there are no type signatures given to either 'printer' or
'printCatalog'.

> Are you saying, that the second error causes the first one? Thus,
> everything is caused by 'printer' being recursively defined?
Yes, that's what I believe. I'm not sure why it shows the first error
message at all; in my experience, strange things happen when ghc
continues after an error message that you shouldn't pay too much
attention to. Perhaps ghc-6.4 just assumes identical contexts in
recursive groups without checking the type signatures.

> Are contexts with differing length in a recursive definition ok in Hugs
> (as in the latest version of ghc) and therefore with Hugs I don't get
> the first message either?
Yes.

Thomas