Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!usc!ucsd!ucsdhub!hp-sdd!apollo!apollo.hp.com!yon
From: yon@apollo.HP.COM (David Yon)
Newsgroups: comp.lang.smalltalk
Subject: Data Hiding...
Message-ID: <4820c7e4.20b6d@apollo.HP.COM>
Date: 19 Jan 90 13:23:00 GMT
Sender: root@apollo.HP.COM
Reply-To: yon@apollo.HP.COM (David Yon)
Followup-To: comp.lang.smalltalk
Organization: Hewlett-Packard Apollo Division - Chelmsford, MA
Lines: 22


	Thanks for all the responses to my posting on Data Hiding
in Smalltalk.

	First off, to those who questioned this...YES, I know *why*
the data-hiding breaks down, that was never in question.

	However, what I wanted to bring to light is that there is
an awful lot of Smalltalk code (at least when browsing the STV/286
system) that just answers one of it's instance variables, instead 
of making a copy.  Although this is standard practice, it would seem
easy to be lulled into a false sense of security, since after all: 
"only an object can modify it's own internal state".  If "get" and "put"
methods merely answer references rather than new instances, this 
is decidedly not that case.  There are of course many contexts in 
which answering with a reference is the right thing to do.  There 
are also many times where it's unclear whether you should answer by
reference or with a new instance.

	Thanks again for the response.

David Yon