[DGD] Fw: Re: callout cantrip

[bart at wotf.org]

On Sat, 14 Mar 2009 04:13:31 -0700, Shentino wrote

More or less, yes.

Just like in your example, I'm changing the array after having passed it to
the call_out, and using the fact that those changes will be visible to the
called out function later.

Bart.

> So you use the callout's arguments as a sort of "floating buffer"?
> 
> On Sat, Mar 14, 2009 at 4:07 AM, <bart at wotf.org> wrote:
> 
> > Bah, as usual, sent this from the wrong email addy :P
> >
> > ---------- Forwarded Message -----------
> > From: "Bart van Leeuwen" <bart at bartsplace.net>
> > To: "All about Dworkin's Game Driver" <dgd at dworkin.nl>
> > Sent: Sat, 14 Mar 2009 12:06:25 +0100
> > Subject: Re: [DGD] callout cantrip
> >
> > On Sat, 14 Mar 2009 02:16:49 -0700, Shentino wrote
> > > Ok, so I was sitting back for a bit, experimenting with callouts,
> > >  and I suddenly had the bright idea of trying to pass a callout its
> > > callout handle as an argument.
> > >
> > > Naturally, owing to causality, not very straightforward.
> > >
> > > Apparently arguments on pending callouts share the object's
> > > dataspace, so what I was able to do was pass a one element array,
> > >  have the callout scheduled, and then modify the array in place ala
> > > TLS/call_trace, thus giving the callout information on what its
> > > handle was. ___________________________________________
> >
> > I've used this trick before for another purpose, it works really well.
> >
> > Lets say you want a daemon like object to distribute events which roughly
> > works as follows:
> >
> > objects can register themselves to the daemon as 'interested' in an event
> > objects can send a notification to the daemon that an event has occured,
> > which
> > will then be sent to all objects that are 'interested'.
> >
> > Now, the obvious solution is to simply loop through all interested objects
> > and
> > call the event notification function in them, but this has a few problems:
> >
> > - This all happens in a single execution round/thread
> > - A runtime error in an event handler will terminate the loop unless
> > caught.
> >
> > So, a slightly more advanced solution might be desirable.
> >
> > The solution I am using is as follows:
> >
> > When the event occurs and the daemon gets called, it starts with doing a (0
> > delay) call_out to itself with as argument an array with a few elements,
> > the
> > 'next' object that shoud be processed, the event this concerns, and
> > possible
> > arguments to the event.
> >
> > Once this call_out is started, the daemon starts a loop that first gets the
> > object from the the array that was also passed to the call_out (lets call
> > this
> > the 'current' object), then determines if/what object should be handled
> > after
> > that one, changes the array accordingly, and then calls the event handler
> > in
> > the 'current' object.
> >
> > Now, it keeps looping there until it is out of time (it monitors ticks
> > used)
> > or until an error occurs. (each time the loop runs it starts with getting
> > the
> > 'current' object from the array and then changing the array to point to the
> > next object before actually calling the event handler)
> >
> > When its out of time or an error occured, the call_out it started earlier
> > will
> > ensure that all interested objects will get the event notification still.
> >
> > As an aside, gurbalib's event daemon is being changed to a somewhat similar
> > implementation.
> >
> > Bart.
> > --
> > Created with Open WebMail at http://www.bartsplace.net/
> > Read my weblog at http://soapbox.bartsplace.net/
> > ------- End of Forwarded Message -------
> >
> >
> > --
> > Created with Open WebMail at http://www.bartsplace.net/
> > Read my weblog at http://soapbox.bartsplace.net/
> >
> > ___________________________________________
> > https://mail.dworkin.nl/mailman/listinfo/dgd
> >
> ___________________________________________
> https://mail.dworkin.nl/mailman/listinfo/dgd

--
Created with Open WebMail at http://www.bartsplace.net/
Read my weblog at http://soapbox.bartsplace.net/