Object >> toFinalizeSend:to:with:

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Object >> toFinalizeSend:to:with:

Henrik Sperre Johansen
The two uses I've seen of this method
(RequiredSelectors>>registerLifelongInterestOf:inAll: in Core, and
AXWeakSubscription in AXAnnouncements on SqueakSource) have used this as
a poor mans replacement for ephemerons.
The problem is it does not work the way these senders expect (ie:
registering an action for each resource to be taken when the receiver is
finalized), but only executes the last action registered for an object.

Would it be an idea to change the implementation to conform to the
expectations of the senders, rather than considering the uses as bugs,
seeing as how that seems to be the common usage? (And a quite more
useful one in the absence of real ephemerons, if you ask me)

Cheers,
Henry

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: Object >> toFinalizeSend:to:with:

Stéphane Ducasse

On Feb 19, 2010, at 10:33 AM, Henrik Johansen wrote:

> The two uses I've seen of this method
> (RequiredSelectors>>registerLifelongInterestOf:inAll: in Core, and
> AXWeakSubscription in AXAnnouncements on SqueakSource) have used this as
> a poor mans replacement for ephemerons.
> The problem is it does not work the way these senders expect (ie:
> registering an action for each resource to be taken when the receiver is
> finalized), but only executes the last action registered for an object.
>
> Would it be an idea to change the implementation to conform to the
> expectations of the senders, rather than considering the uses as bugs,
> seeing as how that seems to be the common usage? (And a quite more
> useful one in the absence of real ephemerons, if you ask me)

Changing for better :)

BTW we should remove the old and not used algorithm left from the browser
supporting traits required methods.
I hope that when adrian is settle down he will give that code a kick. :)

Stef

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project