[racket] TR error message

Yes. This is an HLI problem that we wish to tackle at some 
point.  We have discussed it more than once.  Ideas and 
implementation contributions welcome. -- Matthias



On Sep 22, 2010, at 8:25 AM, Shriram Krishnamurthi wrote:

> Isn't it possible to have a significantly simpler *printed* type even
> if you WERE exploiting this information?  The fact that
> 
>  Complex Complex
> 
> doesn't type is true even if the type of < were "merely"
> 
>  Real Real Real *
> 
> That doesn't preclude < from ALSO being of other types, like
> 
>  Integer Integer
> 
> and so on.  That is, the others are surely subsumed by some more
> general type, and the error message can merely print the "least
> subsumed" type(s) relative to the arguments.
> 
> Shriram