[asterisk-dev] The fate of 16033

[Olle E. Johansson]

18 feb 2010 kl. 02.10 skrev Kirill 'Big K' Katsnelson:

> On 100217 0051, Olle E. Johansson wrote:
>> I would have liked for someone to have added a quick note saying "thanks for contributing! You're on hold" so you know that the bug tracker is still alive...
> 
> I am rather with Russel on that point. Yes, the patch has been marked ready for testing immediately, and I regard that as a confirmation. If the report and the patch slipped through cracks in the attention field after having been ACKed, then that's a real problem.
> 
> This is understandable, as the report inflow is huge. It would be interesting to define a procedure of re-synching: if a bug not taken care of in e. g. a month, how to "ping" to non-intrusively and without any judgement of the report contents but call attention to it.
> 
>> On a side note - to catch attention from the right people you want to add proper information in the subject.
> > If you had said "SIP autocreatepeer issue #16033" I would have read it earlier.
> 
> Thank you, advice appreciated. I have been asking how to get attention of the "right people" like module maintainers etc. -- but got no answers until your recommendation.
> 
> I'll prepare new patches against heads of 1.6.1 and 1.6.2 branches in a day or two. There was a new function call added to destroy an iterator between 1.6.1.6 and .12, and the current patch fails against 1.6.1.12 anyway.
> 
>> I don't think the autocreatepeer code has changed much in many, many years.
> 
> Does not look like it has been used much at all? What I am missing somewhat there is kind of a peer "template" that would be applied to an autocreated peer -- you can configure a manual one in a [section] of sip.conf, but little can be configured in an automatic one.
> 
Funhy you should say that. In my chan_sip2 code over five years ago, there was a template system for autocreatepeer as well as for realtime. Maybe we should bring that template code alive again.

> Do you think such a feature could be of any interest to the community of users? I have not taken a decision to implement it yet, but I'll put that on the "pro" side of the balance.
I think it's needed both for autocreatepeers and for realtime. I often get questions if the templates we have in text based configs is available for realtime, but it is not. We do need something for that.
If I remember correctly, I had something like

[template_autocreatepeer]
type=template

[mysweetrealtimeclass5]
type=template

These where excluded from all matching and counting of objects, but used when creating the peers. I think the implementation would be much more simple now.

/O