[Grace-core] Question from Phil Wadler

[Kim Bruce]

Thinking of our target audience (novices), I don't see much gain for them in exchange for the added complexity.

Kim



On May 20, 2011, at 11:06 AM, Andrew P. Black wrote:

> Ah, I'm beginning to remember.  By not distinguishing methods from fields (they allow updates of both), they make life hard for themselves.  Contravariance gets into field update, for example.   I guess the real question is: now that A+C have done the hard work, is there anything that we can gain from it?
> 
> 	Andrew
> 
> On 20 May 2011, at 08:01 , Kim Bruce wrote:
> 
>> I'm away from home and my notes, but here's a brief (hopefully mainly accurate) response.  The Abadi-Cardelli calculus models a different style of language than represented in most well-known OO languages.  Because they allow updating of methods (not just instance variables) it is much harder to model in the lambda calculus, for example.  (See my paper with Cardelli and Pierce "Comparing Object Encodings" to see the differences in encoding.)
>> 
>> Theorists have tended to follow the Abadi-Cardelli model, while language designers (and programmers) have tended toward class-based models that do not follow the model in Abadi-Cardelli.
>> 
>> Kim
>> 
>> 
>> 
>> On May 20, 2011, at 10:42 AM, Andrew P. Black wrote:
>> 
>>> Phil Wadler asked me today:
>>> 
>>>> Why isn't Grace based on the Abadi-Cardelli calculus?  -- P
>>> 
>>> The truth is, I haven't looked at the A-C calculus for such a long time, that I don't remember.  IIRC, it's an object calculus, which treaties classes as derived rather than primary entities, so to that degree, we are similar.  How they handle inheritance and initialization of mutable objects, I don't recall.
>>> 
>>> 	Andrew
>>> 
>>> _______________________________________________
>>> Grace-core mailing list
>>> Grace-core at cecs.pdx.edu
>>> https://mailhost.cecs.pdx.edu/mailman/listinfo/grace-core
>> 
> 
> _______________________________________________
> Grace-core mailing list
> Grace-core at cecs.pdx.edu
> https://mailhost.cecs.pdx.edu/mailman/listinfo/grace-core