Re: consensus on :query ?

I'd think there would be plenty of sites where this could be a big win
HPACK-wise
where a small number of paths are used with a large number of different
query strings.
Do we have any data one way or the other?  It seems like evaluating this by
data would make sense.


On Sun, Jul 20, 2014 at 3:50 PM, Mark Nottingham <mnot@mnot.net> wrote:

> <https://github.com/http2/http2-spec/issues/556>
>
> I’m not hearing anyone else express interest in this.
>
> Comments? Unless there’s a clamour of support soon, I’m inclined to close
> with no action.
>
> Cheers,
>
>
> On 12 Jul 2014, at 3:19 am, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:
>
> > Some time ago I proposed and argued that :query should be split out
> > from :path and as I perceived it, there were a lot of people nodding
> > on that.
> >
> > I don't want to inject it right now, we have plenty of balls in the
> > air, but can I get you top open an issue for it, so we don't forget ?
> >
> > (Some of) the arguments for :query:
> >
> >       Load balancers and other "triage" proxies seldom if ever
> >       look at the :query part to determine handling, splitting
> >       this field out of :path means they don't have as much data
> >       to run through the decompressor to make their decisions.
> >
> >       Semantically :path is much more static than :query is, which
> >       means that :path can be compressed by back-reference, whereas
> >       :query almost always will need huffman coding.
> >
> > --
> > Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> > phk@FreeBSD.ORG         | TCP/IP since RFC 956
> > FreeBSD committer       | BSD since 4.3-tahoe
> > Never attribute to malice what can adequately be explained by
> incompetence.
>
> --
> Mark Nottingham   http://www.mnot.net/
>
>
>
>
>