home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Collection of Internet
/
Collection of Internet.iso
/
infosrvr
/
doc
/
www_talk.arc
/
000271_davis@dri.cornell.edu _Thu Oct 29 19:57:52 1992.msg
< prev
next >
Wrap
Internet Message Format
|
1992-11-30
|
3KB
Return-Path: <davis@dri.cornell.edu>
Received: from dxmint.cern.ch by nxoc01.cern.ch (NeXT-1.0 (From Sendmail 5.52)/NeXT-2.0)
id AA20826; Thu, 29 Oct 92 19:57:52 MET
Received: by dxmint.cern.ch (dxcern) (5.57/3.14)
id AA19462; Thu, 29 Oct 92 20:09:57 +0100
Received: by willow.tc.cornell.edu id AA17963
(5.65c/IDA-1.4.4 for www-talk@nxoc01.cern.ch); Thu, 29 Oct 1992 14:09:05 -0500
Date: Thu, 29 Oct 1992 14:09:05 -0500
From: Jim Davis <davis@dri.cornell.edu>
Message-Id: <199210291909.AA17963@willow.tc.cornell.edu>
To: www-talk@nxoc01.cern.ch, wais-talk@quake.think.com,
nsb@thumper.bellcore.com
Subject: In-band v.s outband labeling. (was: Misconceptions about MIME [long])
I wish that people would take a less inflammatory tone in
the mail, and try to assume that even if some parties may
be wrong, they are at least possessing good will and a desire
to find solutions that work for all users.
Larry says he has a problem. As I understand it, his problem
is that moving data is expensive, so he wants to have more information
available at the client before retrieving the data. Why would anyone
object to this?
Does it help to consider that someday, maybe soon, much network data will
be sold for a fee. Would you object to customers wanting information
about data before paying the price?
If we agree this is a real problem, then we face the problem
of finding a solution that is practical, efficient, compatible,
and so on.
Now, as N Freed points out, no method exists to automatically
compute in the general case the resource requirements of e.g. PostScript,
that has little bearing. For one thing, there are many data types
which are not turing complete programs (e.g. GIF files). For another,
the person who produced the PostScript in the first place probably
knows the measurements. After all, if I created a .PS file and put
it on the net, I probably displayed it at least once, so I know well
enough that it does halt. (unless I am malicious.) And finally,
even if the creator does not tell you, you can still compute it, once,
and then store the results in the file archive, or Wed, or whatever.
The fact that a full, general and automatic solution is impossible
does it lessen the value of partial solutions.
So the proposed solution is to have more explicit labeling. What's
wrong with that? If you dont use it, it costs nothing.
If you do use it, it makes life easier for people.
You can argue that many people are too lazy to provide this information,
and I agree But not everyone is lazy. there will also be people motivated
to expend the trouble to label the data, yes, even to go to the
extent of computing how much main memory you will need to uncompress
and so on. Why will they be motivated? Because those who do will have larger audiences, and that is already a motivation. And when money starts to
enter the picture, the motivation for market share will be stronger.
There will be a consumer's union and an underwriters laboratory for
information space. The question is where to put the information,
the "meta data" if you will.
Best wishes