[PRL] the "success" of types

[John Clements]

On May 19, 2006, at 12:49 PM, Matthias Felleisen wrote:
>
> 2. There are two major industrial type systems that are safe/sound:  
> Java and C#. These also use parametric polymorphism.

Would it be petty of me to point out that the parametric polymorphism  
in Java (a.k.a. Generics)  can't be used (AFAIK) usefully in  
invariants, only at function calls?

That is, to the best of my knowledge I can't write an invariant that  
states:

Right now, x and y are both lists of some type T (though at other  
times they might not be of the same type).

... whereas I _can_ write

This procedure accepts two lists that must both contain the same type T.


(I'm not sure, with type erasure in Java 1.5, whether I can encode  
the former using the latter. In fact, I'm pretty sure I can't.)




This suggests to me that the "success story" of parametric  
polymorphism in Java can be applied only as long as you eschew the  
imperative style.


John

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2484 bytes
Desc: not available
Url : https://lists.ccs.neu.edu/pipermail/prl/attachments/20060519/b66b6ca6/smime.bin