Set & Table hierarchy

[Mark S. Miller <mark>]

Date: Thu, 16 Nov 89 12:41:55 PST
   From: heh (Hugh Hoover)

   implementor's note on Table->copy()

   I am implementing it with essentially the same semantics as
   t->asImmuTable()->asMuTable().  The copy of a mutable does not share any
   structure with the original table.  

Semantically, it is correct for them not to share structure.  However,
I thought the point of the copy-on-something stuff was so that they
could share structure in a semantics-free way.  If the first
implementation always actually copies, this is fine.  We just need to
be sure that a copy-on-something system is what we are building
towards. 

   However, the individual elements in
   the table are NOT copied, so if they are not stateless, a change to one
   of the elements would show up in both new tables.  I think this is correct.

Absolutely.  To do any different, we would have to make the same
referencing vs containment distinction in X++ that we make in
Orgls&Berts.  Eventually we may want to (to more fully integrate
them), but not for a looong time.