[asterisk-dev] [Code Review] SIP: Pineapple

[Klaus Darilion]

Kevin P. Fleming schrieb:
> Tzafrir Cohen wrote:
> 
>> If there a reasonable chance that a potential broken device that needs
>> such a backward-compatibility[1] option actually diverts from the
>> standard in all fronts? Or would it make sense to split it into several
>> sub-options (technically: make it a bit field)?
> 
> In the 5+ years I've been involved with Asterisk, I cannot remember any
> case where someone enabled pedantic mode and then had to disable it
> because a device was unable to interoperate with Asterisk. With that
> said, though, the vast majority of Asterisk systems never have pedantic
> mode enabled, so that's not a very representative sample of the possible
> endpoint problems we might experience.

I am not sure about current Asterisk, but in the old days, with 
pedantic=no Asterisk accepted out-of-dialog INVITEs with to-tag, which 
is a very bad choice if you just use it as gateway and use a SIP proxy 
which is not dialog aware (eg. transaction stateful only) as in-dialog 
requests (to-tag present) are usually not authenticated by the proxy.

regards
klaus

> Given what 'pedantic' currently controls, I really can't come up with
> any scenarios where an endpoint could expect us to work properly by
> *not* doing the pedantic checking/parsing. All of the pedantic toggles
> just increase Asterisk's RFC compliance.
>