- From: Lucas Pardue <lucaspardue.24.7@gmail.com>
- Date: Thu, 25 Jul 2019 23:48:40 -0400
- To: Willy Tarreau <w@1wt.eu>
- Cc: Brad Lassey <lassey@chromium.org>, HTTP Working Group <ietf-http-wg@w3.org>
- Message-ID: <CALGR9oZ7CyJ3LD4rmJn+4=E83ad3qc93Nc82-uJMXjiRL+NQjA@mail.gmail.com>
Hi Willy! On Thu, 25 Jul 2019, 23:23 Willy Tarreau, <w@1wt.eu> wrote: > Hi Brad, > > On Thu, Jul 25, 2019 at 02:49:52PM -0400, Brad Lassey wrote: > > Hi all, > > Lucas and I put together a draft to capture at least one of the ways > > forward that were identified at the priorities side meeting yesterday > > morning. Please have a look. > > > > https://datatracker.ietf.org/doc/draft-lassey-priority-setting/ > > https://github.com/bslassey/priority-setting > > While I think that's a very wise approach, I think it would have more > success if it doesn't change the default behaviour. I mean, some > developers on both sides have gone through great length implementing > priorities and expecting now that they should suddenly have to add > extra code to continue to use them might be difficult to accept. > Conversely, all those who didn't implement them would have a great > incentive to add a few lines of code to signal their lack of support > and immediately expect a better experience. Thus I think priorities > should remain enabled unless signaled as disabled by this new setting. > The aim is to maintain todays default behavior of endpoints supporting H2. This is achieved by defining the initial value of the setting as 1; endpoints "opt out" by sending 0. Do you think we have mis-specced this compared to our aim? And for having been late to implement them into haproxy (still not > present), I'd definitely adopt this proposal as a temporary measure > if it helps improve user experience! > That's an interesting data point, thanks for sharing. Cheers Lucas
Received on Friday, 26 July 2019 03:49:16 UTC