Having topped up on nutrients we are ready to begin the first session of the afternoon.
The morning session basically covered where things were at last year and updated them based on progress etc.
As the session progresses Leif Madsen has been adding content to the wiki page:
We have some new faces in for the afternoon session. Namely:
Brad Watkins: Compuware
Abel Camarillo: Neuroservices Communications
Nicolas Bouliane: Avencall
Xavier Carcelle: Avencall
David Deaton: Digium
Mathieu De Bellefeuille: Nisse
Alex Balashov: Evariste Systems
Joel Schuweiler: Digium
Shawn McCord: Cycore Systems
Klaus Darilion: ipcom.at
We started by bringing people up to date with what happened this morning.
Alex Balashov brought up SIP path support with an offer to spend some time on it. Klaus Darilion had written the original patch and while it works it doesn't support realtime. Kevin Fleming said that as the backend of AstDB will be able to be stored in a realtime DB, the registration fields will be removed from realtime anyway.
We then actually moved on to discuss the wiki. Do people use it etc? Basically yes. It needs to be updated more. It needs to be more clear how to become a contributing editor. It's been a good platform and allows creation of documentation straight from the source - meaning it keeps in sync.
In the future the integration between Jira and the Wiki etc (such as linking to related bugs) will be added.
The bug tracker supports voting but at this stage it's not being used - there is a danger of disappointing people if there are lots of votes but an issue isn't getting work on.
Alex brought up the overwriting of rtp port number or rather the reusing of port numbers. The idea of doing more efficient RTP port allocations via maybe a linked list of oldest ports used etc. On a loaded system it can be easy to reuse an existing port.
We then discussed config files. The sample config files have a lot of information in them - way too much for the average user. Maybe we should export the available options to the wiki etc. That way you could also have something that parsed the config files and tell you if it's correct.
Return to last known state - if a reload fails return to a previously working state rather than failing. Also being able to just go back to a previous state after a restart rather than the current state. One other option is to use SVN etc to contain your config files.
Another option would be to force Asterisk not to start if a module failed to load. This is mostly usefully in a cluster where there are other machines that can take over.
There was a brief discussion of Async AGI and whether it would be supported in the future - mostly because Adhearsion was thinking of moving entirely to it.
We've decided to have a break for a while till at least 4pm then do a wrap-up.