A problem for "Web as distributed programming paradigm"

Adam L. Beberg beberg@mithral.com
Fri, 24 Aug 2001 10:36:35 -0700 (PDT)


On Thu, 23 Aug 2001, Russell Turpin wrote:

> Jeff Bone asked me to think about the REST stuff,
> and my thinking took me to a rather brute, perhaps
> trivial, realization. (1) Some people are hoping
> that Web protocols evolve into a true distributed
> programming paradigm. (2) Right now, they aren't.
> HTTP and SOAP are both IPC mechanisms.

How about... HTTP POST is now the only way to get packets from machine A to
machine B, thus you have no forkin choice in the matter.

POST is simply a send() and a recv(), with zero or more bytes in each, and
with a slightly fancy destination address because we only have one port
now, 80.

Simple, elegant, and I can do anything I want with this, with very minimal
poking and proding. I can ASCII encode the data first and call it XML, I can
put the send() in the address and call it GET, yadda yadda...

Yea, call it whatever you want, but that's all it is, send+recv in an ever
connection restricted network. HTTP is a system with enough noise that I can
sneak some signal through, and there is the real value. But that's it.

As for distributed programming with it, as long as you can get packets from
A to B to A, the rest is trivial, wait for a packet, do something, and
reply. You can't hide that from the programmer since they have to write the
"want to do something" part and the "do something" part. You can try and
hide the remote part (which is what you want in the rest of your post), but
it will break.

The acronym and language used changes from week to week, but we all know
that's just to keep the hourly rates up and hide the fact that nothing has
been new since the 70's ;) ... There is this wonderful new language with
portable bytecode, no pointers, and tons of neat libraries... it's called
Pascal.

- Adam L. "Duncan" Beberg
  http://www.iit.edu/~beberg/
  beberg@mithral.com