Re: [hybi] Masking only Payload/Extension Data

[Andy Green <[e-mail]>]

On 03/09/2011 07:29 AM, Somebody in the thread at some point said:

> In hope of getting consensus on the idea that only the payload and
> extension data should be masked and not the framing itself, I took a
> pass at adjusting sections 4.1 and 4.2 accordingly.  It didn't take
> much, just a few minor tweaks.
>
> What do you think?  Any chance we could reach a rough consensus on
> masking only the extension/payload data?

+1 from me.

It also solves Toni Ruottu's slightly exotic case where he might want to 
mask his server frames, since he can do that as a legit extension 
negotiated at handshake-time, stick his mask as first extension data, 
then and the framing and content structure is identical to masked client 
frames.

I also remind folks that to set the clear length field in Javascript, 
you have to get the browser to accept to create an object of the 
requisite length.  That extends past 1GB for 4 bytes of ASCII and 256GB 
for 5 bytes of ASCII starting with a letter.  Since the next thing is 
defined to be the mask in extension data, 3 or 4 bytes of control is all 
you can realistically hope for.  If you can knock over an intermediary 
with 4 bytes, then it will fall over anyway once every 4G frames that go 
through it and beyond caring about.

-Andy