[DGD] Any advice?

[Ben Chambers]

----- Original Message -----
From: "Stephen Schmidt" <schmidsj at union.edu>
To: <dgd at list.imaginary.com>
Sent: Monday, May 19, 2003 11:54 PM
Subject: Re: [DGD] Any advice?


> > Ben Chambers <bjchamb at bellsouth.net> wrote:
> > > Any advice as to where to start or what will be
> > > necessary to create a new kernel from scratch?
>
> Having done something similar (I would not label Melville
> a "kernel" for a variety of reasons, but it is a standalone
> mudlib for DGD) the main thing is to make a list of all
> objects the driver needs to access (this used to be only
> two objects, the driver object and user object - it may
> have changed) and all the functions it tries to call
> (quite a few in the driver and I believe receive_message
> is the only one in user, and again that's probably
> changed in the 9 years that have elapsed). Decide what
> you want all those functions to do, and then your
> new kernel is basically set to implement. (Figuring out
> how you want user to handle receive_message is going to
> force you to cope with command processing and most of
> the game-world type issues at some point.)
>
> On Mon, 19 May 2003, Noah Gibbs wrote:
> >   One of your most major decisions is going to be how
> > (and whether) to handle inheritance and upgrade-in-place.
>
> I agree. Melville doesn't really deal with this issue at
> all, because it wasn't a feature when Melville was designed
> nine years ago, and I've never had the time to work through
> the whole thing myself to make Melville handle it better.
> I have done that primarily because I think that if I wanted
> to go down that road I'd probably be better off just making
> a variant of Melville that ran over the kernel. I have
> started work on that project, but at present rate of progress
> it'll take about 250 years to complete it :(
>
> In fact, it's not clear to me why one would want to write
> a new kernel from scratch, except I guess for the sheer
> challenge of it. The kernel mudlib handles a great number
> of very hairy persistance issues and it seems to me that
> the time to code from scratch something to handle them,
> and then debug it, would be prohibitive.
>
> But if the idea is to do it "because it's there" then it
> would be an excellent project for sharpening one's skills.

While the challenge is part of my reason for doing this, I also feel that
coding it myself will allow me to have more flexibility in how it is
implemented and what exactly I want it to do and how it does it.  For
example, I'm sure that because Melville was written from scratch, it is more
like what you originally intended than if you had changed your ideas in
order to allow the mudlib to be built on top of the kernel.
> Steve
>
>
> _________________________________________________________________
> List config page:  http://list.imaginary.com/mailman/listinfo/dgd

_________________________________________________________________
List config page:  http://list.imaginary.com/mailman/listinfo/dgd