[asterisk-biz] Res: Res: 2008 Predictions

[Gregory Boehnlein]

>  A tremendous amount of dev time has been spent in the latter half of
> 2007 making 1.4 'more stable and reliable' because at the developer's
> conference in mid-2007 it was determined that this is where the focus
> needed to be. I've noticed the improvements and a survey taken in
> another thread that happened recently seemed to indicate there were
> others that have as well.

I have been very encouraged to see the massive amount of effort being spent on getting 1.4 "production ready". There are many new features and core changes in the 1.4 branch that introduce an exponential amount of new, possibly untested situations that could cause serious issues in production systems. As an example, all of the race conditions and deadlocks that have been fixed due to the increased focus on using tools like SIPP, as well as Murph's focus on stabilizing AEL, not to mention all of the VLDTMF related issues that have been squashed. The IAX2 changes have also been a sorely needed update as well.

> There were other contributors that also noted areas of improvement
> that were still necessary to get 1.4 where we all want it to be, and
> work still continues in this direction. Many of the astobj2 conversions
> that happened in 1.4 happened because they were desperately needed to
> put in proper reference use counts around objects that needed to be
> protected because some thread somewhere was still using it resulting in
> crashes and mutex deadlocks. You need only subscribe to the svn-commits
> list to see all the activity that still occurs in the 1.4 branch. It's
> not limited just to security fixes.

Yeah.. I love seeing 4-5 patches a day happening to the PBX core, SIP channels, Zaptel and IAX drivers. Those are really the core of what I use Asterisk for, and those changes have really added to the stability of the pre-production boxes that I have in the field. The real litmus test for me will be happening in Q2 2008 when I rebuild some of my TDM gateways using 1.4.
 
> Digium has made a conscious decision to stop patch support on 1.2 in
> favor of focusing efforts around the active 'stable' branch and the
> active 'development' branch. The community members there at the time of
> the dev conference, myself being one of them, fully supported this move
> knowing that even with what Digium already provides there is a finite
> limit to the number of branches that can be supported at any given time
> and the line had to be drawn in the sand somewhere.

I'm in support of this move.. the 1.2 branch is very solid for the specific tasks that I'm using it for. I'm well aware of the issues that it has, and that makes it a solid, mature branch to use for my needs. 

> That being said, this is an open source project. I don't believe
> there's anything at all that stops anyone out there from picking up
> where Digium and others have decided to "leave off" and fork the 1.2
> branch to allow it to stay active with patches and move it forward.