[asterisk-users] Discussion: 1.8 quality issues

[Olle E. Johansson]

28 apr 2011 kl. 16.53 skrev Russell Bryant:

> 
> ----- Original Message -----
>> PS. Please don't start a discussion about 1.8 quality in this thread,
>> that's a separate issue. I just want to know what you think about
>> closing 1.4 support now. If you want to discuss 1.8 quality, start a
>> new thread. Thanks.
> 
> I don't think it's a separate issue at all.  I would like to see discussion of exactly which issues are preventing users from using Asterisk 1.8.  We're trying to shift focus to those issues and get them resolved as quickly and as efficiently as we can so that we can all move forward.
Thanks for ignoring my plea... Please at least change the subject ;-)
> 
> Resources are limited.  What is the best use of our time to help ensure the best future?  Where do we want to see the project in the next 6 months to a year?  A primary focus on further solidifying Asterisk 1.8 is what gets us there in my mind.
I agree.
> 
> Asterisk 1.4 was released 4.5 years ago.  It mostly "just works", and I fully expect many to keep using it until they see a need to migrate.  
If you think it's mostly "just works" it can't be hard to support it a while longer then, can it?

> This process has been likened to when the community moved from Asterisk 1.2 to 1.4.  Asterisk 1.8 has been much more stable out of the gate than 1.4, due to many things we have done over the years to increase quality, including:
> 
> 1) We have adopted peer code reviews as common practice for all non-trivial changes going into Asterisk.  This alone has _greatly_ increased the quality of the code going in.  It is rare that a patch goes up for review where someone doesn't point out some sort of problem.  These problems are found and fixed _much_ faster in the up front review process than if it had been many months later when someone encountered it as a bug in the field.
Agree. But it also puts a significant delay on the process. We have to be very careful about that. Having too many branches open in addition to this was a pain. With fewer branches I hope it will get better.

> 
> 2) We have placed an increased emphasis on automated testing efforts.  In addition to building up a lot of test environments inside of Digium, there is now an open source automated testing effort for Asterisk.  There are over 200 test cases that run every time anyone touches the code.  This includes complex call scenarios such as transfers and call parking.  These open source test cases touch about 25% of the code (and what it does touch are things we considered some of the most important parts).  That is a huge step forward from where we started.  We are continuing to place more and more resources on this effort to move it forward.
Agree. It's great and we need to continue working on it, because it obviously hasn't caught everything we should have caught. I fully agree that it is a wonderful system and I've said that many, many times.
> 
> Despite comments in this thread, there _are_ many people using Asterisk 1.8 in production, including large installations.  The ones with systems working perfectly fine don't tend to make as much noise.  :-)  For those still getting hit by problems, I hope that you can make the time to report them so that we can work with you to get them resolved.
I don't disagree there either. I have only stated that it fails in my and my customer's installations. Everyone is using Asterisk in different ways. If it did not work anywhere I would be very disappointed.
> 
> I started my work on Asterisk as a volunteer 7 years ago and even though it is now my full time job, I still put many personal hours into the project.  I care very deeply about the success of Asterisk.  I truly believe that the steps we have taken with release management are in the best interest of the project.
I understand that you do, I don't think you do things you don't believe in. But you do need feedback from production sites to make the best decisions. 

What you bring up here is important but in my world have no relation to the decision about 1.4. I understand you want to use development resources in a good way, but there are also marketing/business perspectives to consider here. I personally don't think closing 1.4 support today is in the best interest of the project from a marketing point of view, as I don't believe we have a working alternative to offer. I understand we have different opinions about it.

Regards,
/Olle