Re: VT100 etc...

Tim Berners-Lee (timbl)
Tue, 6 Apr 93 21:51:06 MET DST

Lonnie Mandigo sums up the problem well. Looking at
Lonnie's diagram, the top box in the top box looks
like, in our case, the WWW network access library.
In the bottom of the box are the vt100 client and
X client boxes -- why not add Windows and Mac clients?
That is, you don't HAVE to have another line between
the network access bit and the display software, it can
all be linked into the same program. As it is for
most WWW clients.

The interesting bit is the line down from the top
box to the bottom box, and this is especially relevant
when it represents a phone line. Then the top box
becomes a gateway between the internet protocols in all
their glory and a special phone line protocol.
Then, the bottom box (the user's software) and the
top box (gateway or internet-connected client)
are basically the same except with the internet access
code unplugged and the phoneline acess code plugged in.
As you say, the display and access code are separated.

The requirements or the phone line protocol are that
it should be efficient, it should allow long transfers
to be stopped at short notice to make way for new ones,
and it should, preferably, look ahead to guess what the
user might want next, and transfer it while he is reading
what he has got. This would take advantage of a reasonable
amount of cache disk space on the user's machine, if he has
it. The ideal is to keep the phone line humming as it is
the bottleneck. The user browses around, with an apparently
very good response time. All the scrolling and such is done
locally so that is instant.

There is not as far as I know any existing public protocol
which does this. If anyone knows of one, please say!
If anyone wants to [form a group to] define and implement
one, then why not. I see this as an important step
toward getting the internet information world out to
everyone in schools and homes.

Tim Berners-Lee
WWW Project
CERN, Europe