Proposal: Collection Literals

Mark Mahieu markmahieu at
Mon Mar 30 23:49:20 PDT 2009

Are you referring to the idea that a reified collections API may have to be
separate from (and not directly compatible with) the current one?


2009/3/31 Neal Gafter <neal at>

> On Mon, Mar 30, 2009 at 5:45 PM, Joshua Bloch <jjb at> wrote:
> > Could you please be a be a bit more specific?  In particular, could you
> > provide an example of  something that this proposal precludes but the
> > for-each construct (which is already in the language) does not?  Once we
> > have such an example, we'll be able to weigh the pros and cons
> > intelligently.
> Sure.  As the proposed construct is defined to create non-reified
> collection types, and that distinction may be highly visible to
> programmers, programs using the proposed syntax would have to continue
> creating non-reified collections.
> With the for-each loop, on the other hand, the only object created by
> the construct is an iterator.  The iterator is constructed by a
> library method that is completely under control of the Iterable that
> comes in, and hidden from code containing the for-each loop.  While
> non-reified types might continue to produce non-reified iterators,
> reified classes may well produce reified iterators.

More information about the coin-dev mailing list