On Jul 20, 2005, at 21:35, Nicholas Clark wrote:
For now c&p helps :-)
I don't agree. c&p is evil, because it causes code bloat and
propagates bugs.
And I'm not after parameterising the aggregates - I'd like to
instantiate
another instance of what is currently a singleton, for the same ty
On Thu, Jul 14, 2005 at 06:14:52PM +0200, Leopold Toetsch wrote:
> Nicholas Clark wrote:
>
> >I'd like to be able to provide this as a non-singleton PMC class, with the
> >addition of a value lookup, and (possibly) iteration. But cut and paste is
> >evil.
>
> Well, the more general question is:
Nicholas Clark wrote:
I'd like to be able to provide this as a non-singleton PMC class, with the
addition of a value lookup, and (possibly) iteration. But cut and paste is
evil.
Well, the more general question is: how can we parameterize our
aggregates. We already have tons of arrays with (b
Right now ponie is still reference counting. In effect it's duplicating the
work of parrot's DOD registration scheme. For space and efficiency reasons it
would be very useful to move the ponie reference counts outside the actual
PMCs, and effectively store them in a hash, much like the DOD referenc
