[e-lang] CapTP in E progress report #5

[Kevin Reid]

On Aug 13, 2007, at 16:27, David Hopwood wrote:

> What are the precise meanings of "unconnected" and "disconnected"  
> here?

An unconnected ref is something that used to be a far ref (and  
therefore has the same identity as it); a disconnected ref never was  
one.

> Alternatively, where should I have looked to answer this question?

http://www.erights.org/elib/equality/same-ref.html

Note that its claim that "(Due to this behavior of Disconnected  
references, we say that == tests designational equivalence rather  
than computational equivalence.)" is  becoming false, and directly  
related to this discussion:

Originally, if one had a disconnected ref, and then successfully  
reconnected and got a reference to that object again, then the two  
refs would be == but not behave identically.

When I was discussing CapTP with MarkM several weeks ago, I argued  
that this was unacceptable:

   (a) It makes sameness too weak; it is immensely useful for == to  
guarantee that what it considers the same will behave *absolutely  
identically*.

   (b) My new safe proxy-ref system <http://wiki.erights.org/wiki/ 
Proxy> makes it impossible for the comm system to break sameness in  
this way.

-- 
Kevin Reid                            <http://homepage.mac.com/kpreid/>