On 16/12/2013 3:55 a.m., mohamad pen wrote: > On Sat, Dec 14, 2013 at 6:57 PM, Francesco Chemolli wrote: >> >> On 14 Dec 2013, at 16:00, mohamad pen wrote: >> >>> greeting every one. >>> I did some experiments on on both squid 2.7 and squid HEAD.bazaar >>> (from launch pad) which both have claimed to support collapsed >>> forwarding (ability to create on connection to origin for every >>> in-transit instances of a url) and none of em doing it right!!! the >>> squid 2.7 (with collapsed_forwarding on) is claimed to support it >>> from the squid's main site >>> (http://www.squid-cache.org/Doc/config/collapsed_forwarding/) >> >> Hello Mohamad, >> what branch did you try from launchpad? Is it trunk or the one which was suggested by Amos? >> In what ways are your expectations not met? You are not saying that squid doesn't support what you ask for, but that it's not doing it right. What is "right" for you? >> >> Kinkie > > > Hi francesco > the launchpad branch was the one suggested by Amos. the trunk version > does not claim to support collapsed-forwarding. > actually I meant it is not supporting it while it is claimed to > support it at least in 2.6 and 2.7 stable versions and because of that > I said it is not doing it right. it creates different connections for > every requests to static files to origin url while > collapsed-forwarding guarantees just one connection for every requests > for in-transit instance. Um, this is very confusing. Can you please re-word that using the names of things instead of "it"? FYI: "collapsed forwarding" is a Squid custom feature. It is not part of HTTP at all. So "doing it right" is left to personal opinion. Also the 2.6 and 2.7 series were a temporary fork off 2.5 intended simply to collect and test a lot of patches people made against 2.5 C-code when they should have been re-based on the 3.0. Those series were never actually intended to be stable or used for long. The volunteers and sponsors (thank you all!) who are picking up the slack in porting to Squid-3 are fixing a lot of bugs and design problems as they go. So the Squid-3 behaviour is not necessarily going to be identical to the Squid-2 behaviour. Amos
