01-06-27 07.54, skrev [EMAIL PROTECTED] p� [EMAIL PROTECTED] f�ljande:
>
> On 23-Jun-2001 Artur Bergman wrote:
>> I don't see how this is relevant.
> This is relevant as it's a more global way of viewing all the little
> problems we are bumping into. These object interactions should be
> implemented on top of the POE, but we don't have the necessary hooks.
Agreed, but it isn't relevant to the technical discussion.
>> C) The poe kernel should have no concept of the object layer, multiple
>> objet layers must be able to exist. (Just like multiple kernels exist
>> today)
>
> No, multiple kernels can't exist in one process nowadays.
I mean multiple implmentations of kernels. (Event,Tk,Gtk,Native)
>>
>> D) If the poe kernel doesn't manage to transparantly move messages
>> across kernel bounderies the object level will not help, if you don't
>> want to reimplment kernel level code in your object layer.
>
> Herm.... herm... moving messages transparently is harder then it looks.
I agree :) Go go go Leolo
> However, your point about the object layer is well taken. What you are
> saying is that if a Checkout needs a Cart (again with the e-commerce), POE
> shouldn't care. However, if a session goes away, any objects it's using
> should also be cleared out. In a perfect world, a smart _stop handler
> would be sufficient. However, this requires annoying amounts of house
> keeping.
>
> So, let's put all the object relationships elsewhere, and add hooks to POE
> that will tell us when objects "change", w/o object cooperation. By
> "change", I mean create/delete but also "completed some task".
Aren't those hooks supposed to be in the object layer, as POE has no concept
of objects.
>> Multiple layers GCing the same thing is an error prone approach.
> Totaly agreed.
>
> -Philip
>
