Hi Vlad, On Mar 21, 2012, at 12:04 PM, Vlad Paiu wrote:
> Hi Saúl, > > I do not think that this has any implications for strict routers, as if the > previous hop was a strict router, the R-URI would not contain a ;gr > parameter, as the current proxy would not have added the Record-Route header > with a ;gr param, so I don't see any way that a prev. hop strict router can > get treated as a loose router with the new patch. > Right, looks like that approach would work then. > About the other idea, with the adding of a new parameter to the rr module to > disable strict routing support, I don't really see it as more generic, and > frankly I cannot even see how it would work. Because what you want to > accomplish here is to properly route requests that have in the R-URI a GRUU > within your domain. So what would you do, always set that RR parameter ? > Because at startup you can't know if you will generate any GRUUs or not. > Le me elaborate on that: right now the loose_route function (in case of in-dialog requests) will check if the RURI has a local domain in the host part, and if so, it considers that the previous hop was a strict router. I we'd add a setting to disable strict routers support, that check would never be performed and the loose routing rules would always be followed for in-dialog requests. I say it's more generic because it achieves the same goal as if we'd add a explicit check for the ;gr parameter, and it covers a broader topic, which is "assume there are no strict routers in the planet". Regards, -- Saúl Ibarra Corretgé AG Projects _______________________________________________ Devel mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/devel
