Re: Pointer: The Superhighway Steamroller - cern.www.talk #3848

HALLAM-BAKER Phillip (hallam@dxal18.cern.ch)
Mon, 27 Jun 1994 17:42:42 +0200


In article <Cs241n.F2G@news.cern.ch>, hallam@alws.cern.ch (HALLAM-BAKER Phillip) writes:
Xref: CERN.ch cern.www.talk:3848
Newsgroups: cern.www.talk
Path: CERN.ch!alws.cern.ch!hallam
From: hallam@alws.cern.ch (HALLAM-BAKER Phillip)
Subject: Re: Pointer: The Superhighway Steamroller
Message-ID: <Cs241n.F2G@news.cern.ch>
Sender: hallam@dxal18 (HALLAM-BAKER Phillip)
Reply-To: hallam@alws.cern.ch
Organization: Wot!!! Me ????
References: <771F@cernvm.cern.ch>
Distribution: cern
Date: Mon, 27 Jun 94 14:58:35 GMT-1:00
Lines: 61

In article <771F@cernvm.cern.ch>, ses@tipper.oit.unc.edu (Simon E Spero ) writes:

|>Highlights-
|>
|> o The folks at NCSA will be glad to learn that their web
|> server is really running on a CRAY, cunningly disguished as HP
|> workstations (hey Rob - that's cheating :-)
|>
|> o Gopher and HTTP are much less efficient at file transfer than FTP

Demonstrably untrue. For starters FTP requires two TCPIP connections, the
control information is slightly higher for HTTP but FTP cannot handle
compression transparently as HTTP can.

FTP is probably the most braindead protocol ever. The ASCII assumption
causes masses of lost bandwidth every single day because by default FTP
clients transfer in ASCII mode. this means that binariy files are frequently
transfered twice. When both client and server speak ASCII this is more
than a little stupid.

|> o Moguls are plotting to ban ASCII from the networks, in an effort to
|> deprive the masses of 80x24 literature,

Moguls are planning to end the assumption that everyone will communicate in
a Latin character set. ASCII is dead. Only the english speaking world
can use it.

|> o What people really want is 23 copies of Alice in Wonderland on a
|> floppy.

Quite true.

|> o HTML can't be used stand alone.

Untrue. I use it on my Toshiba portable which has no modem and no ethernet.

|> o HTTP over dialup means you have to drop the line for every transfer.

Nope. HTTP 1.0 would require this if implemented in a daft manner. There is
no standard for HTTP over dialup at the moment. But the assumption would be
that you would connect to a proxy server and use a persistent connection.

Hart misses out on the key idea of the Web which is not HTML nor HTTP. The
idea behind the Web is a name that can identify a unique information object
and how to access it. HTML and HTTP are just the icing on the cake.

Note that Prof Hart gives us no solution to the problem of sending the
complete works of shakespere from A to b. The Web does. If you only want
to refer to one verse of hamlet :-

http://shakespere..bogus.rsc.com/plays/hamlet/act2/scene4#verse23

Try that with plaintext.

--
Phillip M. Hallam-Baker

Not Speaking for anyone else.