On Fri, 2008-10-17 at 09:44 -0700, Eric Northup wrote:
> > What we now have in mind is to introduce a new constraint:
> >
> >   (has-field 'a name 'b)
> >   
> I don't see how you would be able to get "protected" fields out of 
> this.  Perhaps that's fine (I've seen it mis-used quite often), but just 
> thought I'd mention it.

1. The C++ notion of public/private/protected is a kludge. Protected
   isn't meaningful in the absence of inheritance, which BitC does not
   propose to have.

2. Private/protected are not true encapsulation, in the sense that the
   types are exposed. This is why #include requirements proliferate in
   C++.

3. It isn't the goal here to support that concept.

> The module system might be a better way to perform that kind of 
> encapsulation anyway.  Is there a module equivalent of "friend"?

That would violate the entire point of a module!


shap

_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to