From: "Patrick J. LoPresti" <patl@skyclad.lcs.mit.edu>
Date: Thu, 20 Jul 95 11:02:41 PDT
To: cypherpunks@toad.com
Subject: RE: Netscape the Big Win
Message-ID: <199507201802.OAA22155@skyclad.lcs.mit.edu>
MIME-Version: 1.0
Content-Type: text/plain


-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "pfarrell" == Pat Farrell <pfarrell@netcom.com> writes:

 pfarrell> Why are all-in-one programs so preferable to using the
 pfarrell> windowing capabilities that are built into every X-window,
 pfarrell> Mac or Windows system?

 pfarrell> Why not use the best mail client, another best webcrawler,
 pfarrell> and yet another news reader?

The problem is that existing operating environments do not, in
general, provide good facilities for the kind of tight integration you
really want.

Besides, there is nothing wrong with a monolithic application, as long
as it provides a sufficiently rich extension language.  Take Emacs,
for example.  Emacs is a monolithic application, but I use different
Lisp packages to read news, handle mail, and develop software.  (I
occasionally even just edit text.)

Moreover, I have co-authored a Lisp package to hook PGP functions into
*every* Emacs mail and news package, without ever talking to the
authors of those packages.  In general, any package can be written to
seamlessly integrate with any other.  In addition, all of these
packages work without modification on every variant of Unix, on VMS,
on Windows NT, and sometimes even on DOS.  I can write in beautiful
(and safe) Lisp, and let the Emacs maintainers worry about the
idiosynchracies of each operating system.

The problem with Netscape currently is that all of their packages are
*built in* by Netscape Communications.  That is why they have no mail
handler, and why their news reader sucks, and why it is impossible for
any of us to fix these things or add a PGP front end.

Java looks somewhat promising; with it, perhaps Netscape can become a
platform-independent system for writing packages to manipulate and
display hypertext.  It would be like an Emacs for hypertext, but with
a crufty extension syntax and no source code.  And a user base 1000
times as large...

 pfarrell> Microsoft has been preaching the use of OLE and component
 pfarrell> programs as its development vision for 2+ years, Macs have
 pfarrell> been popular for ten years, why is the trend still towards
 pfarrell> adding every possible bell and whistle to single programs?

These approaches suffer for two reasons.  First, it's a pain to
incorporate the same basic display code into every package.  Second,
it's a pain to rewrite the same basic display code for every window
system.  (Especially when "every window system" means Microsoft,
Macintosh, and X.)  Other subsystems than display have similar
problems (networking comes to mind), but I think display is the major
pain in the groin.

What Netscape could do is provide the engine for hypertext display,
with a sufficiently rich and simple extension language that it would
be easy to write new modules.  Someone would probably write a decent
news reader.  Someone else would write a mail handling package.
Someone else would write a PGP interface.  And so on.  Netscape would
need to provide other functions across platforms, like TCP sockets,
but that isn't impossible: Emacs has done all of this (save graphical
display) for over a decade.

Gosh, we might find ourselves using 1980 technology by the year 2000.

I don't know enough about Netscape's plans for using Java to know
whether any of this is likely to happen.  I'm not even sure I want to
see it happen.  But it would be interesting.

Cheers.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2
Comment: Processed by Mailcrypt 3.3beta, an Emacs/PGP interface

iQCVAwUBMA6aLHr7ES8bepftAQEIVgP9G8p4lV1+Uc+6cpLZW4hMF+k7CYYp2Jp6
xh0qZXW0Sd7STPn+sP/fGPvErauGTlDiyIoW5bTJ9srITtFN8U1Yr7QollQZPqUa
5Rhbu7LjFTmixpdo0wiDTuUiRObnoE4Pj+/27EiamEqG160TjGiHDyCodh/eyFWS
8+R/yT5RCPw=
=pja4
-----END PGP SIGNATURE-----