Re: [hybi] frame length encoding

[John Tamplin <[e-mail]>]

On Sun, Aug 22, 2010 at 9:53 AM, Shelby Moore <[e-mail]> wrote:

> The more I think about it, I don't think very many routers are going to
> support an open-ended maximum frame size (i.e. 2^63), so the ability to
> use the simplified sendFile() is going to be luck??  But there is no
> reason to not support 63-bit, because there is no one maximum size that is
> going to fit every instance.  The maximum has to be negotiated on
> connection handshake?
>

Most routers aren't going to inspect above the IP layer - they just forward
network-layer packets.  TCP streams essentially have infinite frame length -
does that mean a router can't forward them?

Where it will come into play are the endpoints that must generate/process
the messages and proxies that browsers use to connect to the outside world,
since they will have to store and forward the frames.  That is one of the
reasons that fragmentation is required, so that such a proxy may choose to
dedicate a fixed buffer for each connection coming into it, and send a
fragment as it reaches its buffer size.

Regarding negotiating a maximum frame size, as soon as we finalize the
framing we will start trying to get consensus on the handshake.  I am
certain we can include a maximum frame size in the handshake if consensus is
that it is needed.

-- 
John A. Tamplin
Software Engineer (GWT), Google