2009/1/16 Paulo Pizarro <paulo.piza...@gmail.com>:
>>>> 17.1.1.3: ... The To header field in the ACK MUST equal the To header
>>>> field in the
>>>> response being acknowledged, and therefore will usually differ from
>>>> the To header field in the original request by the addition of the
>>>> tag parameter.
>>>
>>> This refers to ACK to 3XX/4XX/5XX/6XX? Looks like we have a bug there,
>>> I've pushed the fix to darcs.
>
> Not only, but to the 2xx response too. All next requests (ACK/BYE/etc)
> within this dialog must have the same TO header that this final
> response. This is done for backwards compatibility with RFC 2543 for
> dialog identification. According with the third paragraph in 12.2.1.1.

I think 12.2.1.1. talks about original request, not about final
response to initial request? 12.1.2 talsk about setting remote URI
from To header field; however, it does not say whether to use To
header field from response or from request. 12.1.1.1 then mentions
that for 2543-compatibility To header field from original request
should be used.

>>>> - SIP_CC_OE_CE_V_041:
>>>>
>>>> A second final response 2XX with a different Record-Route in the same
>>>> INVITE transaction should change the Route header of the resent ACK.
>>>> Sofia ignores the second Record-Route header and re sends the ACK
>>>> exactly as the first one.
>>>
>>> I think the second paragraph in 13.2.2.4 refers to early dialogs; the
>>> route set should not change after first 2XX to that dialog. (The
>>> wording says otherwise, however.)
>>
>> I agree with you, it's right that the route set should not change
>> after first 2xx to that dialog.
>> A second final response 2XX with a different Record-Route in the same
>> INVITE transaction only makes sense in the fork scenarios.
>> I made a new test with a fork scenario, and sofia did not sent the
>> second ACK with the new route set for the new dialog. It seems like
>> the hairy fork bug :)
>>
>> So, Is it possible to help you to fix this fork bug? Or it is not
>> advisable ... :)

We need more rigorous testing, having those ETSI cases as check test
cases (see check_nua.c) would make me confident. Of course, if you are
not afraid sharp shards of fragile C code, you can have a peek inside
NUA internals...

-- 
Pekka.Pessi mail at nokia.com

------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
Sofia-sip-devel mailing list
Sofia-sip-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sofia-sip-devel

Reply via email to