RE: #540: "jumbo" frames

On Thu, 26 Jun 2014, Gabriel Montenegro wrote:

> NYC interim discussions were very clear about not churning any more before 
> WG LC unless something was completely broken.  We don't think that is the 
> case here, especially as an extension would allow folks to experiment with 
> jumbo frames for data if they so wish.
>
> So I agree with Nicholas, Roberto, Mike on this one.

As do I.

This discussion seems to mostly reiterate what we've already discussed 
numerous times. I think it is more important to get a protocol together that 
interops than to bikeshed forever on tiny details - and yes I think this is a 
rather tiny detail all things considered as I believe we all can live with 
what we have right now - even if it hurts a little bit more for some. And I 
think muxing and prio are two fundamental parts of the protocol that large 
frames don't really play nicely with.

Let me also remind everyone that at least two browsers only will speak HTTP/2 
over TLS and lots of the talk in this thread seem (to me) to assume plain-text 
implementations.

-- 

  / daniel.haxx.se

Received on Thursday, 26 June 2014 02:23:15 UTC