[Asterisk-Users] Re: Asterisk forking, Was: Digium Website Update: Asterisk Business Edition

[Lee Howard]

Andrew Kohlsmith wrote:

>On Saturday 11 June 2005 19:51, Lee Howard wrote:
>  
>
>>I don't think that "lack of mindshare" completely defines the reasons
>>behind Asterisk fork failures.  It places all of the blame on the
>>forkers.  I think the truth, though, is that they not only fail due to
>>"lack of mindshare" but also due to competition from Digium's own
>>Asterisk community.  Forks are not succeeding, yes, but Digium has a
>>hand in that... of course they do.
>>    
>>
>
>I'm not saying you're wrong, but I'm curious: how does Digium have a hand in a 
>fork failing?
>  
>

That's what I tried to explain in my last post, in particular after this 
first statement.  Forks enter a "hostile competition" rather than a 
"healthy competition".

>>I've heard more talk about Asterisk forks than I've ever heard about
>>forks of any other other open-source project.  I think that this says
>>something about how difficult-to-swallow Digium's dual-license decree is
>>for a lot of prospective contributors/developers.
>>    
>>
>
>I disagree; if it were that hard to swallow the project would either be 90% 
>digium-written (it's not) or it would be a total flop (again it's not).
>

If you (or someone else reading this post) is in a position to give 
statistics on what percentage of the code is Digium-written (or 
Digium-rewritten - in the case where a disclaimer is not obtained for 
some unpatented work and Digium rewrites the work independently) then I 
would be thrilled to see it.

>>We see this happen all of the time with the Linux kernel.  It happens
>>with HylaFAX.  It happened with X.  I'm sure it happens a lot with many
>>other open-source software projects.  It happens easily and usually is a
>>"healthy" process because the playing field is even.
>>    
>>
>
>Agreed.   But where are the successful Asterisk forks?
>  
>

I don't know of any successful Asterisk forks (unless 
http://www.asteriskwin32.com is considered "successful" - although I'll 
admit that I'm not really in-the-know).  But this was my point: that the 
way things were set up by Digium makes a successful fork difficult.  
Digium always has an upper-hand, and things were set up intentionally 
this way.  Again, I don't take particular issue with this.  I'm just 
trying to explain why forking Asterisk would not be a particularly easy 
task.

>>Of course, this "healthy" forking cannot be done with Asterisk because
>>Digium will not accept any non-disclaimed code into their repository.
>>    
>>
>
>... What you'd described about distribution-maintained patches has nothing to 
>do with this.  Digium could take a distribution-maintained patch and rewrite 
>it into Asterisk proper under the dual license (as could any other 
>contributor) and you'd still gain the benefit of the patch.  I'm not sure I 
>see where you're going here.
>  
>

If you (or someone else reading this) has the necessary information to 
provide statistics on how what percentage of the code comes from 
rewrites of non-disclaimed code, then I would be particularly interested 
in hearing it.  I suspect, though, that it is a rather small - perhaps 
insignificant - amount.  But, yes, providing that there is not a patent 
involved - yes, the work could be rewritten and integrated.  But this 
was my point: that given the right environment forks can benefit from 
each other.

The one thing that an Asterisk fork can never do, though, is relicense 
itself.  Only Diguim can do that.  If Digium had wanted an equal footing 
in this regard then Asterisk would be LGPL or BSD or something a bit 
more liberal.  So if I'm a manufacturer of PBXes and have some 
proprietary IP that I do not wish to be GPLed, then if I want to use 
Asterisk somehow, then I can really only work with Digium for 
licensing.  All of the other forks will be license-prohibitive.

>I have to admit that I know quite a few people with their own 
>modules and such to replace what they feel is bad code and just won't 
>contribute it back to Asterisk due to the friction they've received about the 
>patch.  I, on the other hand, tend to bitch loud and continuously enough and 
>wear them down to the point of accepting it.  :-)
>  
>

So we're not in disagreement, it would seem.  Getting code contributions 
into Digium's Asterisk codebase is not something that many average 
people are going to want to undergo.  From what I've seen, "friction" is 
a bit light of a term for it.  It seems much more hostile than that.  
And, that's often even before the disclaimer hurdle is reached.

Lee.