I guess it's because just about all other parameters have a value, so people wrote their parsers to expect ";some_parameter=some_value" and then when they came across ";lr", they thought, "my parser won't like this, so I'll just stick a value in there".
Of course, that's pure speculation. But anyway, the practice of lr=anything is wrong. Unfortunately, this incorrect behaviour has been so widely-implemented, SIP equipment now has to interpret ";lr=1", ";lr=on" and ";lr=true" (and ;lr=anyvalue) to all mean ";lr" otherwise you get interop problems. Regards, Attila -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Iñaki Baz Castillo Sent: 26 November 2008 21:42 To: [email protected] Subject: [Sip-implementors] Why so many implementations expect "; lr=on" or ";lr=yes" instead of just "; lr"? Hi, AFAIK loose routing is exclusively defined in RFC 3261, and it's clear and explained in *all* the examples and BNF section that "lr" has no value: lr-param = "lr" <sip:proxy1>,<sip:proxy2>,<sip:proxy3;lr>,<sip:proxy4> So I wonder why there are various SIP devices that only do loose route if the Route/Record-Route header has a parameter like: ;lr=on ;lr=yes Of course it's a vendor issue, but... why is this error so common? any historic reason I miss? Thanks. -- Iñaki Baz Castillo _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
