[asterisk-dev] Bugs/patches 16033 and 16590 ignored forever

[Leif Madsen]

Kirill 'Big K' Katsnelson wrote:
> Your explanation is indeed sensible, but there are different resource 
> pools involved in debugging/fixing the problem, and in testing the fix. 
> If I report a bug without sending in a fix, there is a chance that it 
> will be fixed by a developer. Then I confirm the fix, and it is good to 
> go. On the other hand, when I fix the problem by myself, there is no one 
> out there to try the fix, and, therefore, assuming rare enough a 
> problem, the fix will unlikely be confirmed and committed. There are 
> just many more developers willing to fix the bug that does not affect 
> them personally that testers willing to test a fix to a bug that does 
> not affect them personally.
> 
> Such a disposition looks quite paradoxical to me.

I don't quite see it that way though. From the viewpoint of the Digium developer 
resource pool, issues that have patches actually get rated higher on the list of 
issues to be resolved because the engineering effort to resolve such an issue is 
(typically) much lower. With a patch that resolves the issue for the reporter, 
the developer is that much further ahead than if they just had debugging 
information. With the patch, the developer knows the exact area of code to start 
working in, and can potentially reproduce the issue more quickly.

Sometimes it just requires a code review because the issue is that much more 
transparent when a patch is provided.

Of course if the issue does not affect a large number of users, then it may be 
possible an issue could sit for a long time even with a patch. But as Kevin 
states, there are open source developers who hang out in #asterisk-dev who may 
be willing to move an issue forward, and if a patch is provided, it certainly 
makes it easier for them to do that.

I guess what I'm saying is don't stop submitting issues and providing patches 
with the thought that issues without patches get worked on first, because that 
is certainly not the case. In some cases they do, but it generally is because 
they affect a large(r) pool of users.

> (I know -- I have already be advised to find a sympathetic soul or pay 
> someone for testing -- but that does not resolve the paradox: I do not 
> have to do all that If I do not send in a patch).
> 
> I think that Paul Belanger describes essentially the same problem,  but 
> from a different vantage point.

I'm not sure how to resolve the paradox you're speaking of. I'm not even sure I 
agree a paradox exists. Separate pools of resources do not exist for testing and 
development.

The Digium development team does both testing and development, along with code 
review and committing. We don't have separate testing teams outside of the 
development group for each of those tasks. Each issue that is assigned to a 
developer is taken through all steps of the process (reproducing the issue, 
resolving the issue, and closing the issue).

All that can be done is to pick the issues that affect the most number of people 
and work on larger issues (i.e. many development hours that otherwise might not 
be done by the community because development of Asterisk is not their primary 
responsibility). And with the knowledge that not every issue can be resolved by 
this pool alone, it is up to the community to determine what other issues are 
most important to them to resolve.

Now with that said, if you have a particular issue that isn't overly large that 
you need a code review done on to move it forward, Kevin mentioned that several 
developers hang out in #asterisk-dev on irc.freenode.net that may be able to 
move the issue forward. Beyond that, I'm all ears for how we can continue to 
refine the process to move issues forward at a more efficient pace with the 
available resources.

Leif Madsen -- Bug Marshal