MOO-cows Mailing List Archive
[Prev][Next][Index][Thread]
Re: [DISCUSSION] a core
-
Date: Tue, 31 Oct 1995 19:26:39 PST
-
From: Michael Brundage <anomaly@u.washington.edu>
-
Cc: moo-cows@parc.xerox.com
-
Content-Length: 2001
-
Content-Transfer-Encoding: 7bit
-
Content-Type: text/plain; charset=US-ASCII
-
In-Reply-To: <m0tA4eP-000KlDC@foley.ripco.com> from "ThwartedEfforts" at Oct 30, 95 04:34:00 pm
>Interesting point... although the point of a core is to be 'generic', and exist
>as a jumping off point for bigger and better things. A core is just the
>basics -- all the things you think you might need to get a job done. What
>that job is is not up to the core. I don't expect the core to do everything
>I want... what would be the fun in that?
Personally, I would like to have a core that follows the spirit of the
excellent system Dave Van Buren has on AstroVR, only even more so:
Have a generic core that does the things that are more-or-less basic to all
MOOs. Then around this central System Core, have subsidiary cores to handle
specialized tasks, e.g., Mathematica Core to handle interactions with an
external Mathematica server, Help Core, FTP Core, Mail Core, NNTP Core,
Gopher Core, HTML Core, Client Core, Web Core, etc., etc. Each Core object
would in turn point to associated objects (Database, Utilities, Documentation,
etc.) and this object cluster would be in charge of handling all the
associated functionality.
These subsidiary cores could be made highly modular, and a transporter system
could be created for importing/exporting the modules from one MOO to another.
(The ability to analyze the inheritance structure that I've been talking
about would be extremely useful for handling the obvious difficulties that
arise in such a system, since you would need a way to know what builtin
functions are available, what dependencies exist, and so forth, to have
any hopes of making an efficient automatic transporter sytem.)
This would then be a hybridization between an object-oriented approach and
a function-oriented approach, and would allow one to plug-and-play,
rather than scrounge-around-and-hope-I-can-make-it-work-in-my-MOO like
we currently do. It would also, IMHO, make MOO more scalable and portable.
Certainly, large MOOs are out there (Lambda alone is gigantic), but
exporting/importing code from one MOO to the next is anything but standardized.
michael
Home |
Subject Index |
Thread Index