Re: Stream limits draft posted

On Mon, Nov 13, 2023 at 12:30 PM Rory Hewitt <rory.hewitt@gmail.com> wrote:

> > bemasc@meta wrote:
> > If ALPS is in use, I think a new ALPN ID is unnecessary.
>
> *Is* ALPS in use? My understanding is that it's just another expired
> Internet Draft -
> https://www.ietf.org/archive/id/draft-vvv-httpbis-alps-01.html.
>
> Which is clearly a shame, but I'm still (for other reasons) trying to find
> out whether it's been implemented by some vendors and to what extent it's
> supported.
>

We ended up deploying an experimental version of ALPS to solve some
problems of this shape. The standardization half was sadly parked. Partly
this was due to people moving around (I'm less involved in the original
motivating use case for us these days), and partly because, at the time,
the HTTPWG wasn't focused on problems that required solving the early
SETTINGS problem. But the design is sound, and I think, if you want to
solve the early SETTINGS problem, this is the right shape of solution.
Certainly it's far more viable than the alternative Rube Goldberg design.

Between QPACK versions and this rapid reset attack, it sounds like the
early SETTINGS problem is a bit more at the forefront now, in which case
the draft can be revived if that's where we want to go. Don't read much
into the draft expiry.

David