Re: [hybi] WS framing alternative

[Pieter Hintjens <ph at imatix.com>]

On Wed, Nov 4, 2009 at 12:10 AM, Greg Wilkins <gregw at webtide.com> wrote:

> BWTP is currently a thought experiment with lots of possible outcomes...

Let's assume the desirable outcome is (c), a real protocol usefully
solving real problems.

What I've missed for both BWTP and for WebSocket are clear statements
of those problems.  Undocumented assumptions.  Greg, let's collect
this for BWTP to start with: a solid list of the problems we are
aiming to solve.  Then, we can examine alternative solutions, and that
gives us the argumentation to justify BWTP, and for implementors to
understand the cost/benefit tradeoffs that went into every decision.

If it's any help this is how I designed AMQP,
http://www.openamq.org/doc:amqp-background.  That document never went
into the final spec but it was essential to position the protocol
properly, and to defend design decisions afterwards.

(Sadly with AMQP we lacked any kind of diverse community and so I made
some _terrible_ design decisions, such as encoding low-volume commands
in binary, but at least they were coherently argued terrible design
decisions.)

Cheers,
Pieter