Re: Local program execution in WWW browsers

Bert Bos (
Fri, 15 Apr 1994 15:01:44 +0100 (METDST)

Chris Lilley wrote (reformatted a bit, BB):

|Yechezkal-Shimon Gutfreund wrote:
|[ about mosaic applets]
|I see nothing in this that is especially NCSA Mosaic specific,
|although the tome of it suggests that the author is thinking
|exclusively in terms of NCSA Mosaic for X.
|Could I sugest that the few remaining places where X resources and
|so on are mentioned be revised ( alternatives are given anyway) so
|it applies to all versions of Mosaic; and further that it be made a
|general interface so that an applet written to the spec will also
|work with other conforming browsers on whatever platform?
|I feel this would then become a very valuable proposal which I
|suspect would be supported by browser implementors as it would
|reduce the burden on them from feature requests.

A valuable suggestion, but an ambitious one. When I think of
interprocess communication I cannot help but think in terms of Unix
concepts (or X). To abstract away from that and still come up with
something that works efficiently everywhere will be difficult.

Nevertheless, I am going to give it a try (I was going to, anyway),
based on the idea of a protocol that alternates commands and
responses, of variable length up to a certain maximum, and with room
for later adding new commands that won't break older versions.

It's probably too early to attempt a universal solution, but if there
is interest, I'd be happy to occasionally report on my progress.

So, who suggests the first batch of things that a `WWW control center'
and an accessory would want to exchange?


                    / _   Bert Bos <>   |
           ()       |/ \  Alfa-informatica,            |
            \       |\_/  Rijksuniversiteit Groningen  |
             \_____/|     Postbus 716                  |
                    |     9700 AS GRONINGEN            |
                    |     Nederland                    |
                    | |