Re: h2 priority

Hi Pat,

On Tue, Sep 02, 2014 at 11:25:38AM -0400, Patrick McManus wrote:
> On Mon, Sep 1, 2014 at 3:57 AM, Poul-Henning Kamp <phk@phk.freebsd.dk>
> wrote:
> 
> >
> > the intent to ship it to customers on a rushed schedule.
> >
> 
> I disagree that 2.5 years in committee is rushed by any definition. Some
> folks can't be bothered to generate running code or sometimes even comment
> for much of that time; but that pattern persists in any group no matter the
> term or the people (I too am guilty at times.)

To be fair, there hava always be about two camps here :
  - those who could progress fast because they already had a running SPDY
    implementation that they could start from (and fortunately we had them) ;

  - those who had never implemented SPDY and for whom even participating in
    a way to bring some *value* is extremely hard because it's impossible
    to provide data nor to experiment, so they can only run aside the fast
    moving train trying to catch it.

I've experienced this period with some great frustration due to that but at
least I'm happy that the WG has made significant progress. However I would
have loved to have more chances of being active. I know can't blame anyone
for this except myself. But I also know that by the time it takes to completely
change an architecture to adapt to a new proposal (est. 1yr), it will have
completely changed to a point that such low-level architecture will be
obsolete again :-/

Thus I mean it's natural that when starting with an existing design, all
participants do not catch up at the same rate depending on the effort that's
required for them to have something to play with.

Just my two cents. Now I'm back to my work trying to rearchitect haproxy
to support multiplexed streams :-/

Willy

Received on Tuesday, 2 September 2014 16:13:23 UTC