Re: [hybi] Addressing Minimal Events

[Frank Salim <[e-mail]>]

I think that connection sharing and coordinating amongst tabs is well
served by workers, as Ian pointed out. I do not think it should affect
the protocol design.

> What happens when the Javascript
> for the first tab/window is closed and it's acting for all the
> others?  Messy.

I believe this is addressed by workers.

> Unfortunately it does require a lot of code on top of WebSocket to do
> the kind of updates I'm doing.  (Ironic as the point is network
> efficiency and minimal updates).

Code is not bytes on the wire! Addressing your use case in the
protocol makes everyone pay for the features you need, and the cost is
bandwidth instead of code. It is better to address them at a higher
layer with another protocol (AMQP? BEEP?). Likewise, the multiple
widget case is best served by an application protocol with
publish/subscribe semantics.

> From the above, my guess is WebSockets look compelling for the "3
> widgets on a page need updates" case, and for the "I like writing my
> own protocols and having full control" case.
>
> But they don't look compelling for "I want to write an attractive web
> application with lots of updating documents and tiny widgets without
> using a pre-existing application framework or large bit of
> Javascript".

I think that this is correct. However, I think that there is nothing
wrong with using JavaScript to implement an application protocol. This
will be reusable code that is cached in CDNs. It will be much like
using a library that implements a protocol in C.

-Frank