Re: [hybi] Masked framing VS mask in frame

[Andy Green <[e-mail]>]

On 03/01/2011 12:25 AM, Somebody in the thread at some point said:

Hi -

> Personally, it seems dangerous to not mask the extension data, since
> you don't know what extensions might carry data controllable by an

Yeah that was the context of the example packet, showing that moving the 
mask to be extension data you could control at most four bytes but then...

''To think about controllability of extension data, which would allow 
carrying on the data in the length bytes, leads to your other point 
about if structure of extension data is really OK as it stands''

> Even if we come up with something like a tag/value approach, I think
> we can't ever have intermediaries that understand the payload without
> understanding extensions -- for example, what do they do when they see
> a reserved opcode after an extension they didn't recognize was
> negotiated?

What would typically go in this per-packet extension area?  Maybe it 
could be moved into being just normal payload in other packets with 
extended opcodes, eliminating the extension area concept.

-Andy