Thanks to everyone who has given this consideration. In case people want 
to try it and don't know what were talking about, here's a changeset 
that switches to no-copy and includes a comment. Let us know if you find 
any problems.

Chris Muller wrote:

>>I often use copy in default access methods. The motivation is increaed 
>>robustness; it can be very risky to modify a collection.
>>    
>>
>
>While a read-only accessor may have its place, in *this* case, robustness is
>decreased, not increased.  If you don't want to allow direct access to a
>Collection, then what do you think about providing various enumerating / adding
>/ removing / finding api from the containing class?
>
>I generally disagree with having code that "cross-checks" potentially incorrect
>code elsewhere.  Besides the aforementioned performance degradation, it can
>inhibit learning because it allows misuse and then even increases
>confusion/uncertainty about the necessity of it when you try to remedy it (just
>like right now, we're all scratching our heads about this!).
>  
>
I agree; there are lots of ways to screw up a Morph, copy or no copy. 
Also, exporting the actual object reduces the need to make lots of 
collection lookalike accessor methods to avoid a performance penalty.

Thanks,
Eddie

-------------- next part --------------
A non-text attachment was scrubbed...
Name: NoCopySubmorphs-efc.2.cs.gz
Type: application/octet-stream
Size: 377 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20050728/cdf40828/NoCopySubmorphs-efc.2.cs.obj