home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Nebula 2
/
Nebula Two.iso
/
SourceCode
/
MiscKit1.7.1
/
MiscKitArchive.mbox
/
mbox
/
000165_misckit-reques…aska.et.byu.edu_Wed Apr 6 12:41:34 1994.msg
< prev
next >
Wrap
Internet Message Format
|
1994-10-30
|
3KB
Return-Path: <misckit-request@alaska.et.byu.edu>
Received: from alaska.et.byu.edu by darth.byu.edu (NX5.67d/NX3.0M)
id AA11378; Wed, 6 Apr 94 12:41:19 -0600
Received: from YVAX2.BYU.EDU by alaska.et.byu.edu; Wed, 6 Apr 1994 10:53:26 -0600
Received: from DIRECTORY-DAEMON by yvax.byu.edu (PMDF V4.3-7 #4169)
id <01HAUOVWF0XCR5NACH@yvax.byu.edu>; Wed, 6 Apr 1994 10:53:09 MDT
Received: from gac.edu (nic.gac.edu) by yvax.byu.edu (PMDF V4.3-7 #4169)
id <01HAUOVNYROGHSQ4QT@yvax.byu.edu>; Wed, 6 Apr 1994 10:53:00 MDT
Received: by gac.edu (NeXT-1.0 (From Sendmail 5.52)/GAC-1.0.R9309061)
id AA04537; Wed, 6 Apr 94 11:52:01 CDT
Date: Wed, 06 Apr 1994 11:52:01 -0500 (CDT)
From: kane@gac.edu (Chris Kane)
Subject: Re: Comments on MiscAgent
To: zet@cip.e-technik.uni-erlangen.de
Cc: misckit@byu.edu
Message-Id: <9404061652.AA04537@gac.edu>
Content-Transfer-Encoding: 7BIT
Juergen Zeller <zet@cip.e-technik.uni-erlangen.de> writes:
> As Christopher pointed out, MiscAgent would provide a easy to
> learn framework for any common Application.
Well, not an *entire* framework, not everything. But another
piece. Not, in itself, for instance, a document controller system.
A document controller system has demands placed on it that I
haven't envisioned MiscAgent satisfying. It's directed more
towards (as you say later) the miscellaneous user-interface
windows and panels and related classes. The model of panels==actors
(in the real-world sense) and MiscAgent==talent agency works
well, I think.
> A looking ahead design of the MiscAgent will nearly be impossible
> due to the Kit's very nature. A MiscAgent as a quite informal
> framework could lead to a problem like the one stated above:
> Nobody knows exactly what's happening and - even more important -
> the exact point where it happens can hardly be found by a new or
> different developer.
Again, I think that framework is a bit too strong of a term for
this. The root and manager of a heirarchy of classes dedicated to
a particular purpose, perhaps. And a very generic one--an application
would contain an instance of MiscAgent, but I don't expect that
messages would be sent directly to it by the programmer.
Documenting the capabilities of classes managed by the MiscAgent
is the responsibility of the programmer and/or writers of the
specification and design of an application. The problem described
(confusion in a project's development) seem to have resulted from a
lack of software engineering methods in the specification and design
phases of the project, not from the particular functionality of
any piece of the software.
And on my original note:
Another possibility, perhaps better than forcing agents to be
subclasses (at some level) of MiscAgent, would be to have a protocol
which is conformed to by classes/objects desiring to be "managed".
This begins to hint at a wider scope for Misc*Agent than class
and .nib loading and message forwarding, though, which is not my
intent at this time.
My hope would be that people would think about what sort of features
such a thing as I described would have, and how it would behave,
and perhaps make lists of their thoughts. No one or two people
can think of everything (not in the short term, at least :)).
Such lists can be emailed to me, if you find you have something you
want to share.
Christopher Kane
kane@gac.edu