Re: HTTP/2 extensions and proxies

Variation on it. From what I saw, you said:

"If we need end-to-end extensions, then intermediaries need to ignore,
but pass, unknown stuff. However, I think that we need to allow for
intermediaries that don't want to assume the risks associated with
extensions and allow intermediaries to drop unknown frames, but only
if they drop *all* unknown frames."

The language in this I don't like is "drop unknown frames". It's not
clear what "drop" means.

What I'm saying is, you either pass the unknown end-to-end frames
through or send an RST_STREAM... meaning, if you encounter an unknown
frame on a stream and you don't intend to pass it through, you drop
the entire stream, and not just that frame.

- James

On Tue, Oct 1, 2013 at 9:08 PM, Martin Thomson <> wrote:
> On 1 October 2013 12:22, James M Snell <> wrote:
>> Either let everything thru,
>> or let nothing thru. Period.
> That's exactly what I said, isn't it?

Received on Wednesday, 2 October 2013 04:17:15 UTC