[Grace-core] talking generics with Gilad over lunch

James Noble kjx at ecs.vuw.ac.nz
Tue Nov 5 13:05:03 PST 2013


> Remember our goal: a simple language for teaching, not so far out of the mainstream that it will scare the horses, I mean the instructors!

indeed.    I still have nightmares remembering SIGCSE.  Well flashbacks anyway.
 
> I think that this argues for wanting a special syntax for types that's ugly like other languages, just so that we can be ugly like other languages. But there is no reason that we can't translate that language into core Grace.

so *that's* a nice spin, and I think even Michael will approve :-)

note that we're only talking about syntax for generic type parameters on requests - not the whole shemozzle. 

> Types in Emerald were just objects — but objects with a particular signature.  There was a special syntax for creating such objects (the sub-language of types), and types created with this syntax were guaranteed to be decidable.   But you could, in principle, create other types out of whole cloth, and membership in such types might be undecidable.   This is quite like the idea that all Grace types are patterns, but not all patterns are types.

very much so.

> I think that we need to separate the researchy stuff on types from the simple language that we use for teaching.    If we can't do this, then we need to work more on the extension mechanisms.

yep.   well that is part of this - if we go to generalised messages, do we still need special support for type parameters.  Thinking of this as syntatic sugar (hell, even passing it as a keyword named "type" :-)  is one way around that, lowering the conceptual weight of the special stuff

J


More information about the Grace-core mailing list