Re: PRIORITY extension

On Mon, Jul 14, 2014 at 4:00 PM, Jason Greene <jason.greene@redhat.com>
wrote:

>
> On Jul 14, 2014, at 5:09 PM, Roberto Peon <grmocg@gmail.com> wrote:
>
> >
> >
> >
> > On Mon, Jul 14, 2014 at 1:30 PM, Poul-Henning Kamp <phk@phk.freebsd.dk>
> wrote:
> > In message <
> CAOdDvNr0PQFWm8qg7oz1tmAS3qaJK9O8fWkqoUJR9sqP+RwX1g@mail.gmail.com>,
> Patrick McManus
> > writes:
> >
> > >> I keep hearing this argument over and over. Is the goal just to
> finish, no
> > >> matter what?
> > >
> > >Of course not. But finishing is critically important.
> >
> > Absolutely, but the quality of the result is far more important
> > than some arbitrary deadline.
> >
> > The most important part is deployment.
> > This is true of transports and application-layer stuff.
> > .. and we know it works when folks implement things properly, because we
> have experience in production with real, real-world use.
> > The risk of getting it terribly wrong is low given this implementation
> experience.
>
> What about the stream dependencies though? I assume you mean the
> experience is the simpler SPDY priority?
>
> The spec would be a hell of a lot simpler without this dep complexity,
> which is what I think scares people away (well that and the scary rfc
> comment).
>
>
We know that getting priority right is necessary to experiencing the
potential benefit of multiplexing.
We know that the simply priority scheme from SPDY kinda worked, but was
failling for a number of usecases.
Given that the old priority system expresses a subset of what the new one
allows, it is unlikely we'd be making anything worse, though there is a
complexity price to pay.
-=R


> --
> Jason T. Greene
> WildFly Lead / JBoss EAP Platform Architect
> JBoss, a division of Red Hat
>
>