[asterisk-dev] sip/rtp jitterbuffer in 1.4? (Chicken or the egg?)

[Andrew Kohlsmith]

On Monday 29 May 2006 13:02, Dan Austin wrote:
> Steve's comments about the difficulty of testing code/features that
> have been put in their own branch ring very true.  I was pretty excited
> to see the packetization patchset get its own branch, but the reality
> set in and eight weeks without a comment or commit.

This is why my personal thoughts on svn branches are as follows:

If you want to add a new feature to asterisk, get a branch started, get your 
code mostly working (no crashes, no screwy build problems, little interaction 
with other parts of the code) and then convince Digium to add the feature to 
trunk.  

Trunk is supposed to be the development branch, but it should not be 
continuously broken.  In fact it should almost never be broken.  By getting 
your feature beta-quality stable in your own branch, you can "stay out of our 
hair" until you're ready to have people pound away at it.  Merge it into 
trunk and let all of us who are insane enough to run trunk on production 
systems hammer away at it.  Since it mostly works already we will be able to 
really help you out.

The RTP jitter buffer would have been a perfect candidate for this kind of 
development.

-A.