[Grace-core] Duck TYping at PoPL

[James Noble]

> I've just listened to this presentation at PoPL.  I think that it's relevant to Grace, but don't pretend to understand the technical details.

Hmm, I don't understand the technical details either - 
but I don't think we want to start out with a dependent type system with an SMT solver...
(or, having gatecrashed the SPJ video the grad students were watching, perhaps we do)

> See http://cseweb.ucsd.edu/~rchugh/research/nested/
> 
> system d :: nested refinement types
> 
> Nested Refinements: A Logic for Duck Typing
> Ravi Chugh, Patrick M. Rondon, and Ranjit Jhala
> POPL 2012 [ .pdf | tech report (arxiv) | .bib ] 

the real issue is with this kind of work is about the flexibility
of the object model --- the constraints (or lack of 'em).  One
obvious feature is adding fields after the object is defined, this
system handles it but Grace doesn't permit it. 

James