[544] in Coldmud discussion meeting

root meeting help first previous next next in chain last in chain last

ToDo, or not ToDo.

daemon@ATHENA.MIT.EDU (Sun Nov 6 17:05:36 1994 )

To: coldstuff@MIT.EDU
Date: Mon, 7 Nov 1994 02:59:34 -0500 (EST)
Cc: root@colinm.ozemail.com.au ()
From: colinm@colinm.ozemail.com.au (Colin McCormack)
Reply-To: colinm@colinm.ozemail.com.au

In general, I would rather see nothing done to the server, 
than see a bunch of ill considered hacks thrown into the server,
in the name of I don't know... Progress?  Convenience?
Notional improvement?

The changes suggested have ranged over: cosmetic (however uglifying)
to fundamental (changing method semantics) to soundly reasoned
(of which I've already implemented some, having thought of them
initially :)

There's nothing wrong with talking about them, nothing at all,
but I'd counsel conservatism in action... remain focused on what
it is that a change would enable a core to do, and what a core
can't do unless it gets the facility you're after... don't get hung
up on what, remember `why'.

There's nothing so damaging as an ill considered change,
something that half-solves a problem, or that solves the
wrong problem... it obscures elegant design.

En passant:  regexps should be taken out of String.

On another matter:  Chaos has indicated that he's still into
evolutionary hacking of the server, and I think that's good.

I've gone quite a way to rewriting in C++, am pausing for an
extended think and regroup.
I may even scratch what I have and rewrite again, 3rd time,
before completing.  It would be a mistake to see mine
as a major push to a new architecture, right now:
I'm simply after a different/better implementation,
not so much a modification to the C-- level interface,
which I consider (in overall conception) to be excellent.
So there's no impediment to getting on with cores, is there?

(`the more I look at it, the more I like it.  I do
think it's good.  No matter how I turn it, or which way I
view it, it remains consistent.  I WISH YOU WERE HERE TO
SEE IT!' [Indiscipline - King Crimson])

Richelieu (Get an easier name to spell) has indicated an
interest in evolutionarily hacking the server too.
I see no reason why they can't work together, if they
want to; or apart, if they must.
Certainly, also, Chaos and Crag would be a moderating interface
between Richelieu and I, to the extent that I might want to
make my position/feelings felt on some issues.

There will always be someone pushing the envelope, and delivering
the goods.  My money's on Chaos, who produced workable tasking
when it appeared that tasking was necessary, and because, somehow,
I trust him not to pollute coldmud with poisoned confectionary 
like do {abomination} while {abominable}.

As far as I'm concerned, you +could+ treat it as an open market,
with people sending out mutually incompatible servers:
To the extent that a core is needlessly based on a spurious
feature, it's not going to be taken up.  To the extent that a
really good new feature is presented by one, it will be adopted
by the others... There's some efficiency in people all working to 
common goals, but let's face it:  the core is what most people
see, and the server's merely a means to that end.  If Rich thinks he's
making it easier to make a better core, and he and I disagree,
more strength to his arm... time will tell.

All you really need is a place to ftp the latest server(s) from
and a way to tell which is what, and the sense to choose and combine
appropriately, so long as the developments are appropriately modular.

If someone hijacks the whole concept, and makes of it an abortion,
so what?  We still have 0.12.0, and its capabilities are far
from exhausted.

The scrabbling for official position is a scrabbling for
the right to apply the name `coldmud'.  I really don't care what
the system I'm working with is called, so long as I can call it
`the best.'  In other words:  I'd hope that the coldmud 
community had the sense to see that whoever's speaking most
sense at any given moment is speaking for coldmud's future. 
If not, that's where we part company, and coldmud departs from
any future it may have.  The good news is:  there is no absolute
zero, it's a partial order.

If you have a problem, articulate it first as a problem,
not as a solution.  The real gems in any development history 
are the problems solved in arrival, not the fully formed
solutions, because those problems permit many possible solutions.
If you have a problem with the way coldmud works, or doesn't,
phrase it as such, so Riche's `method parsing method' might be
something like:  `I wish there was some way for an object to
accept an arbitrary method call / message, because it'd enable
me to do something I couldn't otherwise, and I reckon this is
a good way to do it' followed by a solution sketch.

As I see it, the problem is mainly that people have difficulty
getting the known-latest server... so someone fer chrissakes
decide where they're going to deposit servers and cores,
stick to it, and get on with it.  Of course, I could be wrong...
there could be other problems.  For all I know, there aren't many
problems, and people are just floating ideas they'd like to
see experimented with... so, grab a server and start hacking,
don't assume that because someone else isn't doing the hacking
for you, but rather chasing their own daemons, that they
don't speak for coldmud, and you do.

For the moment, I'm core-extending, as I know now where (my)
coldmud's headed, I can make the right choices.  Server and
core are codependant, in a relation of mutual presupposition.
I have a feeling it's time to push the core envelope, some.


My A$0.02.