Re: [hybi] WS framing alternative

[Greg Wilkins <[e-mail]>]

Jamie Lokier wrote:
> Let's not get confused about what will and won't work.
> 
> It doesn't matter whether WHATWG sanctions any new frame type or not:
> you can't _use_ a frame type if the other end won't recognise it.
> 
> That's because the message boundaries are determined by recognising
> the frame type byte, then parsing the message in a type-specific way.


Is this true?

The high order byte is currently defined as selecting between
sentinel or length encoding, and the 7 lower order bits are ignored.

So as the spec is written - regardless of the value passed, it
will be one of those 2 framing types.

In essence, the byte is 1 bit framing type and 7 bits undefined.
The 1 bit is also indicating content-type (or at least content-encoding),
which is a strange mix of purposes.

I hope I'm right because if what you say is true, a server might
eventually have to implement up to 256 different framing types!

... and as much as I'm able to endlessly debate the benefits of
one framing type over another, I really don't care that much.
Can't we just pick one and use it!   If the sentinel framing
was good enough to carry anything, then I could live with it.
But it's not, so let's just do length encoding.

Note also that i think it is very valuable to have a known
framing type, so that intermediaries can know what message
boundaries are and treat them appropriately
either with forwarding data, logging, debugging etc.


regards