RE: feedback on draft-nottingham-httpbis-alt-svc

I would prefer that Alt-Svc is a separate document.  I think that this is a large add to the base spec where we are trying to close down issues.  I know it is not popular, but I would also be favorable to revisiting the previous discussions about extensibility, again with the caveat that there is a simple model which would not require a major overhaul of the spec.

Regardless of where it is documented, Alt-Svc needs to be optional.  This is both for Alt-Svc as a whole and for the various pieces like TLS for HTTP URIs.

Thanks!!

-Rob

-----Original Message-----
From: Martin Thomson [mailto:martin.thomson@gmail.com] 
Sent: Thursday, March 6, 2014 1:41 PM
To: Roberto Peon
Cc: Mark Nottingham; Julian F. Reschke; HTTP Working Group
Subject: Re: feedback on draft-nottingham-httpbis-alt-svc

On 6 March 2014 11:48, Roberto Peon <grmocg@gmail.com> wrote:
> So, it depends on the schedule for the document as compared to http/2 
> and how we tie up other aspects, I guess..

Unless we want to rewind on some of the decisions we've made, I only see two options:

1. Make Alt-Svc a normative reference of HTTP/2, effectively including its contents by reference, much as we are doing with HPACK.
2. Inline Alt-Svc.

In both cases, we don't get to publish HTTP/2 until Alt-Svc is done.

I know that we are trying to ensure that inlining doesn't happen, but I think that in this case, there is an advantage to it.

Received on Friday, 7 March 2014 09:23:22 UTC