On Fri, Apr 3, 2009 at 3:40 PM, Benny Prijono <bennylp at teluu.com> wrote:
> On Thu, Apr 2, 2009 at 11:46 PM, Alexei Kuznetsov <eofster at gmail.com> wrote:
>>
>> Hi,
>>
>> Some SIP phones produce Contact like "user at domain@x.x.x.x". pjsip fails to
>> parse such headers:
>>
>> 17:24:24.363 sip_transport. ?Error processing 800 bytes packet from UDP
>> 10.0.0.1:5060 : PJSIP syntax error exception when parsing '' header on line
>> 11 col 36:
>> SIP/2.0 200 OK
>> Via: SIP/2.0/UDP 10.0.0.24:55187;rport=55187;branch=111
>> From: "John Smith" <sip:john@xxxxxxxxxxx>;tag=222
>> To: "John Smith" <sip:john at example.com>;tag=333
>> Call-ID: 444
>> CSeq: 555 REGISTER
>> Expires: 300
>> Contact: <sip:john at 10.0.0.24:55187>;expires=300
>> Contact: <sip:john at 10.0.0.24:55175>;expires=139
>> Contact: <sip:john at 10.0.0.24:55134>;expires=4
>> Contact: <sip:john at example.com@10.0.0.41:5070>;q=0.90;expires=31
>>
>
> That's not valid according to the spec. Character "@" is not allowed in the
> user part of URI, so it should have been escaped. FYI
>
> SIP-URI????????? =? "sip:" [ userinfo ] hostport
> ??????????????????? uri-parameters [ headers ]
> userinfo???????? =? ( user / telephone-subscriber ) [ ":" password ] "@"
> user???????????? =? 1*( unreserved / escaped / user-unreserved )
> user-unreserved? =? "&" / "=" / "+" / "$" / "," / ";" / "?" / "/"
> unreserved?????? =? alphanum / mark
> mark???????????? =? "-" / "_" / "." / "!" / "~" / "*" / "'"
> ???????????????????? / "(" / ")"
Oh yes, I completely understand that. I'm just trying to survive in
such cases. Other SIP phone can register with the wrong Contact on the
other computer, but with the same credentials. And pjsip will fail to
do its job without even cooperating with that bad SIP phone directly.
What we get is a failure to register just because some other phone
registered with the bad Contact somewhere. I just thought it might be
reasonable for pjsip to survive in such situations and not to be so
dependent on that. What do you think?
Alexei
