Re: [EXTERNAL] I-D Action: draft-ietf-quic-transport-26.txt

[Matt Joras <[e-mail]>]

+1. Facebook has a similar intention. We currently have a variant of
draft-24 (with a different QUIC version number) widely deployed and have
been waiting on the (hopefully) last wire format change which is now in
-27. After that I am hoping we will be on that version for the foreseeable
future as a stable version.

Matt Joras


On Fri, Feb 21, 2020, 2:02 PM Ian Swett <ianswett=
[e-mail]> wrote:

> As you saw, -27 was released because it did not include #3169.
>
> The intent of Google and Chrome is to implement and ship -27 as widely
> as possible.  There are a number of design issues out for consensus call,
> but most should not have significant impact upon deployments and none are
> wire format changes.
>
> If the WG would like to declare -27 as a more stable deployment draft,
> that sounds great to me, since we intend to operate as though it is unless
> something drastic arises.
>
> Ian
>
>
>
> On Fri, Feb 21, 2020 at 4:08 PM Praveen Balasubramanian <pravb=
> [e-mail]> wrote:
>
>> It makes sense to include any TP format change and publish draft-27
>> immediately. But independent of that I'd like the WG to declare the
>> intended stable draft version. Maybe at the IETF 107 meeting?
>>
>> -----Original Message-----
>> From: Christian Huitema <[e-mail]>
>> Sent: Friday, February 21, 2020 9:08 AM
>> To: Praveen Balasubramanian <[e-mail]>; [e-mail]
>> Subject: Re: [EXTERNAL] I-D Action: draft-ietf-quic-transport-26.txt
>>
>>
>> On 2/21/2020 8:57 AM, Christian Huitema wrote:
>> > On 2/21/2020 8:31 AM, Praveen Balasubramanian wrote:
>> >> Draft-26 just dropped with minor changes. The planning discussion in
>> IETF 106 was to stabilize and settle on a draft version for six months or
>> so to get deployments and data. It'd be nice for implementers to get some
>> idea of what that version is supposed to be. I was assuming it'd be
>> draft-25 but that's clearly not the case. Are there any more changes that
>> are a must have for wide deployment? I think only major security issues or
>> deployment blockers would qualify.
>> > There are  open 41 PRs, 15 of which are marked "design" . In
>> > particular, PR #3169 did not make it into the draft. That PR would
>> > change the format of transport parameters. There was a rough consensus
>> > to adopt it in Zurich (I was in the rough). Since we don't see it in
>> > draft-26, shall we conclude that it was abandoned? Or is "stabilizing
>> > for 6 months" just one of those things that people say but don't really
>> mean?
>>
>> On the same topic, there is a poll in the implementers' Slack chat.
>> Results so far are 12 votes to "abandon draft-26 and release draft-27
>> immediately", 0 against. There seem to be little appetite to rev the
>> implementations for minor changes, and a willingness to test the new
>> transport parameter format asap.
>>
>> -- Christian Huitema
>>
>>
>>