Richard A. O'Keefe wrote:
> NEW RESULTS!
>
> There are people in this mailing list who are of the opinion
> that the quirk in the implementation of #removeAll: is not a bug
> but an obvious precondition that programmers should always have been
> aware of, so it doesn't need fixing.
>
> I am deeply indebted to mwai at geico.com for telling me that
>
> 	|coll|
> 	coll := OrderedCollection with: 1 with: 2.
> 	coll removeAll: coll
> 	coll inspect --> OrderedCollection()
>
> _works_ in IBM Smalltalk, despite IBM Smalltalk having the same
> implementation of #removeAll:.

Finding a single class in another implementation that just happens to
work in the case you are interested in proves nothing.  You pluck IBM's
OrderedCollection out of the air because it happens to meet your needs.
We are talking about the general Collection algorithms, not just as they
apply to the behaviour of OrderedCollection.  Do all the other
collection classes in IBM Smalltalk do this?  I suspect not.

-Dave