From: gotoken@... (GOTO Kentaro)
Date: 2000-06-20T01:53:58+09:00
Subject: [ruby-talk:03524] Re: Deep copy?

In message "[ruby-talk:03522] Re: Deep copy?"
    on 00/06/19, Hugh Sasse Staff Elec Eng <hgs@dmu.ac.uk> writes:
>Probably, but it would be slow.  I think it is interesting that deep copy
>is not available, though I can see why overuse would be a problem.
>
>I would suggest that it is not the sort of thing the programmer should
>have to manage if the language implememntation could handle it correctly.

A serious problem is the meaning of copy.  More precisely, we don't
have a consensus to what an object is,i.e. the state of an object. For
example, the behavior of an IO object in the future is undecidable, so
a hash including an IO is undecidable too.  Because such binding is
dynamically changing in languages like Ruby, we have to understand the
state of an object more clearly.  This is, unfortunately, known as a
difficult mathematical problem.  Some solutions have been proposed but
a clincher has not came yet. 

-- gotoken