Ivan Zhakov wrote on Wed, Jun 26, 2013 at 02:21:05 +0400:
> On Wed, Jun 26, 2013 at 2:12 AM, Daniel Shahaf <danielsh_at_elego.de> wrote:
> > Johan Corveleyn wrote on Tue, Jun 25, 2013 at 23:22:12 +0200:
> >> On Tue, Jun 25, 2013 at 11:03 PM, Ivan Zhakov <ivan_at_visualsvn.com> wrote:
> >> > On Tue, Jun 25, 2013 at 10:45 PM, Greg Stein <gstein_at_gmail.com> wrote:
> >> >> On Tue, Jun 25, 2013 at 11:55 AM, Philip Martin
> >> >> <philip.martin_at_wandisco.com> wrote:
> >> >>> Branko Čibej <brane_at_wandisco.com> writes:
> >> >>>
> >> >>>> I'm really not a fan of this config knob. Anyone who carries their
> >> >>>> laptop around will effectively have to set this as the default, because
> >> >>>> you never know when the next weird proxy will pop up in front of your
> >> >>>> server. And disabling chunked requests by default is a lot worse than
> >> >>>> the extra non-pipelined request for broken proxies, IMO.
> >> >>
> >> >> Right.
> >> >>
> >> >> Though I suspect most of the problems are reverse proxies in front of
> >> >> a particular server, so you can put the config option into a [server]
> >> >> config block instead of global. That will help to limit the problem,
> >> >> but lack of dynamic detection is still a problem.
> >> >>
> >> > What is the benefit of dynamic detection enabled by some knob in config file?
> >>
> >> The dynamic detection has a cost (1 extra request per connection),
> >> that you might want to avoid by default (most environments won't need
> >> the dynamic detection (especially corporate environments)). Only
> >> enable the dynamic detection if you know the proxy has a problem with
> >> chunkness, or if you're not sure it will stay that way, or ...
> >>
> >> (not interfering with the rest of the discussion right now :-)
> >
> > AIUI the cost is only incurred by set-ups that have the so-called
> > "busted" proxies. And a config option has a cost too: it would need to
> > be supported until 2.0 (aka, indefinitely).
> Please note that this extra request is per session and currently we
> create many sessions even during one operation. And I'm also not happy
> to make performance worse for users who doesn't use reverse proxies
> and etc.
Please define "etc".
Also, I just said that the cost is only incurred only by people who use
a "so-called 'busted' proxy". If you think that is not true, please say
that explicitly, I don't want to have to fish from your words whether
you think that is the case or not.
(We have enough bad implications on IRC right now; don't need more on list)
Received on 2013-06-26 00:25:23 CEST