LispOS: LispEnv or Tunes?
Wed, 30 Apr 1997 12:02:55 -0700
> 3. This web server SilkOS idea (cool name) seems to be just a LispM with HTTP
> support. It's not much different (if at all) from what you can already do today
> with existing Lisp products. Unfortunately, all by itself, it has no ability
> to distribute processing to the client, or to support mobile agents between
Glad you like the name :)
You're pretty much correct that we can do most of this TODAY.
Isn't that a wonderful thing!?
We can pretty much start creating useful things today while we work
to further improve the underlying technology to support it.
> like to borrow their approach to attaining Ubiquity (i.e. a browser-embeddable
> VM) and perhaps support real-time translation of their bytecodes into the
> native wordcode format of a Lisp VM (this as a non-core feature). That's very
> different from the Lisp-on-JVM approach.
OK, but that is probably even crazier than running on JVM ! ;)
It is an excellent idea, but you've just shrunken up the
people who can use your software by a massive amount.
Perhaps I should say (I think it is public knowledge)
that Franz did this already.
We have Lisp running inside Netscape browsers. Who cares? Nobody!
In my opinion it would be a mistake to depend on Lisp running in
a browser because we can do (almost) everything without it.
Now, don't get me wrong. I can see down the road, AFTER we've been
successful just running on the server side, to expand that success
onto the desktop. And infact, in my broader vision, I see that
the code we write on the server can relatively easily be broken
up into components that can also run in the browser.
So instead of all URL accesses (i.e. mouse clicks) going back to
the server, some might stay local. But the point is the software
is written essentially the same! So we can stay just running on the
server until we've been successful and gained enough market muscle
to offer Lisp-enabled client-side processing.
But even before this Lisp utopia has been achieved, we have the
intermediate step of running on existing JVM,
*if it turns out to be truly useful*, which I'm not convinced.
I'll also mention that Microsoft has also done something similiar (i think)
to what I'm talking about. They have some alpha/beta/vaporware thing
called "dynamic html", in which I believe html can be dynamically generated
within the client itself. They are figuring out how to bypass the JVM!