[WF-General] fruits in trees

Alistair Riddoch alriddoch at googlemail.com
Thu Mar 29 12:00:10 PST 2007

On 28/03/07, James Turner <zakalawe at mac.com> wrote:
> On 27 Mar 2007, at 17:20, Erik Hjortsberg wrote:
> This leads to a larger question on how to proceed with Indri/Cyphesis. This
> is obviously a world rule thing, so all work we do on world rules in
> cyphesis has to be redone in indri once that's operational.
> So how far from operational is indri? How much will you be able to put into
> development the coming months?
> In essence: where do we go from here?
> Assuming I am able to make a large-scale contribution is wildly optimistic -
> various factors have greatly reduced my free time, unfortunately. If anyone
> wished to adopt the indri codebase, it can (possibly with some small work)
> be made to run a world, where sear (and presumably ember) can connect, run
> around, talk and use tools. The major hurdle I stalled on is ODE
> integration, which is necessary to provide real physics (movement) and
> collision detection.
> However, Alriddoch has made many incremental changes to cyphesis which (as I
> understand it) make it suitable for larger world deployments than was
> previously the case - i.e I think it will be some considerable time before
> cyphesis is the limiting factor in world construction.
> And as we've said many times in the past, if you have rules written for
> world objects in cyphesis, porting them over is a trivial task, especially
> if your object hierarchy is well thought-out. I would be dismayed if anyone
> was reluctant to write new objects for cyphesis simply because indri is
> theoretically coming along - at the moment it's clearly not.

Thanks for taking time to comment James. I understand your career has
now taken a direction that for the time being won't leave you much
time for hacking on WorldForge, and have tried to plan things
accordingly. My main focus has been to use cyphesis as a platform for
developing techniques and protocol that can be used on any server. The
move to encoding key entity behavior in entity attributes means that
the same rules can be used on a new server implementation unmodified.
The just has to implement the behaviors described by those properties.

Cyphesis now scales quite well, and will get better. It may well not
be the platform for really large scale worlds in the future, but my
current job gives me unprecedented opportunity to learn how really
large scalable reliable systems are built, and I have plenty of ideas
floating around.

I really don't think we need to worry too much at this stage, as long
as we keep in mind that what we do should be fundamentally portable to
a new platform. In my next post, which more directly addresses the
topic of the thread, hopefully I will express how I think this can be

Alistair Riddoch
alriddoch at zepler.org

More information about the General mailing list