[asterisk-dev] chan_sip marked as Extended support in Asterisk 13+

[Paul Belanger]

On Fri, Aug 8, 2014 at 1:54 PM, Matthew Jordan <mjordan at digium.com> wrote:
> Hey everyone!
>
> As you may have noticed, we've branched Asterisk 13 and will soon be
> releasing the first beta release. Asterisk 13 is the next Long Term
> Support (LTS) release of Asterisk, building heavily on the new
> architecture and features developed in Asterisk 12. One of those new
> features is, of course, the PJSIP stack in Asterisk. Today, we decided
> to make a big commitment to the ongoing development of the PJSIP stack
> in Asterisk by marking chan_sip as Extended support.
>
> What does being in Extended support mean?
>
> First, Extended support modules are still supported in Asterisk - just
> by the community as a whole, and not by Digium specifically. A large
> number of modules are in Extended support, and are actively supported
> by members of the Asterisk Developer Community. When security issues
> are found in Extended support modules, all developers - including
> those at Digium - focus on fixing them. When architectural changes
> affect Extended support modules, all developers - including those at
> Digium - do their best to maintain Extended support modules. Some
> developers at Digium also actively participate in maintaining and
> developing Extended support modules, albeit usually in their free
> time. However, such modules are not focused on by Digium developers,
> and issues reported against Extended support modules are generally
> handled by the community.
>
> That being said, chan_sip is Core supported in Asterisk 1.8, Asterisk
> 11, and Asterisk 12. Bugs reported against chan_sip in versions of
> Asterisk derived from those branches will continue to be triaged and
> handled by Digium as well as by the overall developer community. All
> fixes for bugs found in chan_sip will continue to be merged into all
> supported branches. While chan_sip may be in Extended support in
> Asterisk 13, this does not change the support obligation that Digium
> has with chan_sip in those branches where it is marked as Core
> supported. As such, the bug-fix policy for chan_sip is the same for
> Asterisk 1.8, 11, and 12 while those branches remain under active
> bug-fix maintenance.
>
> However, it does mean that when Asterisk 11 enters into security fix
> only mode in 2016 [1], developers at Digium will no longer focus on
> fixing bugs in chan_sip.
>
> So why mark chan_sip as Extended support now?
>
> One of the major motivating factors for replacing chan_sip was the
> support burden we faced with that channel driver. Due to its
> architecture (or lack thereof), changes in chan_sip often have
> unforeseen negative repercussions. Fixing a bug in chan_sip often
> creates a new one. Adding a new feature is difficult, particularly if
> that feature interacts with the SIP protocol at a low level. And,
> while the addition of over 100 unit and functional tests has helped
> tremendously, chan_sip does its best to resist change. I went into
> more detail on the motivation behind moving away from chan_sip in a No
> Jitter blog post [3]; without rehashing the details here, it suffices
> to say that despite the best efforts of many smart, talented
> developers, chan_sip sucks a lot of time and energy from those who try
> to maintain it.
>
> The Asterisk team at Digium consists of approximately ten developers
> and a community support manager (Hi Rusty!). We always attempt to
> focus our development efforts on the parts of Asterisk that give the
> most benefit for the entire Asterisk community. In Asterisk 12, this
> involved writing not just a new SIP stack, but also substantial
> development in the Asterisk core and its interfaces. Focussing on
> improving the fundamental architecture of Asterisk has great long term
> benefits for the community, but it does come at a cost. To make these
> kinds of architectural improvements requires a large investment of
> time and energy. While we were able to do this in Asterisk 12, we do
> not feel we can continue to make these kinds of investments while
> indefinitely supporting chan_sip at the same level as the PJSIP stack.
> There's no question that we will continue to support chan_sip for the
> next two years, but marking chan_sip as Extended support now provides
> some timeframe under which we can plan to devote more resources to
> making bigger and better changes in Asterisk.
>
> So, does this mean that chan_pjsip has the same level of functionality
> as chan_sip?
>
> No. There are things that chan_sip can do that chan_pjsip cannot do
> currently (CCSS, for example). There may always be some features in
> chan_sip that are not replicated in chan_pjsip - and that's okay. One
> of the great things about open source software is that if a feature is
> truly needed, someone will write a patch for it!
>
> So should I plan to move everything to chan_pjsip now?
>
> We have gotten some great feedback on chan_pjsip (thanks SchmoozeCom
> and FreePBX!), and are actively fixing bugs, addressing
> interoperability concerns, and adding new features. We feel that
> chan_pjsip is in great shape and will work for many deployments and
> scenarios. At the same time, it has not been deployed as widely as
> chan_sip, and so there may be interoperability problems that will
> arise. We'll do our best to respond to those issues and get them
> fixed.
>
> Regardless of any criticism that may be levied against chan_sip, it
> has one obvious advantage: it has been around a long time and a lot of
> very talented developers have put countless hours into it. As a
> result, it has accumulated years of interoperability patches and
> fixes, which has made it able to work with a large variety of
> equipment, phones, ITSPs, and other SIP flavored things that I can't
> even predict. The benefit of having so much energy devoted to it
> cannot be underestimated. This is why we worked hard to make sure that
> both chan_sip and chan_pjsip can operate together. If, while using
> chan_pjsip, an interoperability bug is discovered, a user of Asterisk
> has the option to fall back to chan_sip while the developers work on a
> fix. The support status of chan_sip is independent of this scenario
> and is merely a reflection of where we, as developers at Digium who
> spend our careers working on Asterisk, want to focus our effort and
> time.
>
> Hopefully, this e-mail provides some transparency into why we made
> this decision. Asterisk 13 is the first LTS release built on the
> architecture and PJSIP stack introduced in Asterisk 12, and as such,
> we are focused on making sure that users of Asterisk have the best
> experience possible. We would not have gotten here without the hard
> work of everyone in the community: so as always, thank you for your
> continued support of the Asterisk project!
>
*grabs popcorn and prepares for action*

Joking aside, I don't see an issue with this.  Even though I have done
zero testing with chan_pjsip at the moment, I am happy to see it being
promoted.  Obviously there is going to be some bumps in the road
however I'd rather see development efforts focused on chan_pjsip then
split with chan_sip.

If you are a digium developer, find me at Astricon, there will be a
beer in it for you.

-- 
Paul Belanger | PolyBeacon, Inc.
Jabber: paul.belanger at polybeacon.com | IRC: pabelanger (Freenode)
Github: https://github.com/pabelanger | Twitter: https://twitter.com/pabelanger