[OOTB-hive] Useful Add-ons and Patches WAS Re: OOTB contribution issues
richard.esplin at alfresco.com
Mon Aug 1 18:33:07 BST 2016
I appreciate Andreas pointing out ALF-20508. I find it interesting that it hasn't been reported by anyone else.
We have previously discussed our Triage process, as documented in the wiki (https://wiki.alfresco.com/wiki/Issue_Triage_Process). I think it is worth pointing out why this process is structured in this way.
I bulk closed 800 issues last Friday. I only closed issues that were reported before 5.0.d. We still have 1100 non-customer issues to review. We get about two new issue reports every business day, and the rate has been increasing. Most of those reports are not real issues, but it takes a significant amount of time to make that determination. Actually fixing a real issue takes a lot more time.
Instead of reviewing thousands of old issues, we want to spend most of our time implementing new functionality and addressing issues that are blocking successful adoption. We need help identifying which of the old issues are worth investigating further, as the reporter has either found a work-around or moved to a different product.
I believe the system is working. We now know that the 20 reopened issues are important for us to look at again. Even though we expect more issues to be reopened, that saves us looking at over 700 issues. We recognize that there are legitimate problems in those 700 issues, but we do not think reviewing them is the most efficient allocation of our efforts.
This is the third time I have done a bulk close of ALF issues. I have been doing a bulk close after each family of GA releases (after 4.2, 5.0, and now 5.1). Though we are getting better at acting on the issues as they come in, there will always be issues that we don't address in a reasonable amount of time. It is better for them to drop off our list than for us to continue working on stuff no one cares about anymore.
We read every issue, and try to reproduce most of them. In addition, members of our team engage with you as our open source collaborators. We also compare public issue reports with problems reported by customers. As a result, we address a lot of the painful concerns that are raised. Every release includes a lot of fixes to issues that are reported by non-customers through the public issue tracker, and these fixes are often guided by patches submitted by the public. That is why last autumn I was able to justify an increase in the resources we spend on this effort.
We have plans to further improve by making it easier to incorporate the patches that you submit. But that process will never be free, as we have a QA responsibility for every change we make to the software.
We will continue to evolve this process to engage more and to accept more contributions. But we expect that will result in more issue reports coming our way. We accept that the product will never be perfect and we will not be able to deal with every issue in a reasonable amount of time, so we will eventually bulk close many issues.
It is easy for someone to complain that we are not focused enough on his or her particular concern. Those complaints often fail to acknowledge the concerns that we have previously addressed, or the thousands of other people who have concerns that we are also working to solve. It is easier for people to give us their problems than for us to solve those problems. We will never keep pace.
Instead of getting discouraged, we focus on providing a product that people love even after they acknowledge ways in which it can be improved.
On Monday, August 1, 2016 4:22:54 PM MDT Andreas Steffan wrote:
> I am sure ALF-20508 is more of an exception than the rule.
> Still, look at it for a minute and imagine how this appears to the other
> end or to the public.
> And seriously: I think an engineer looking at it should be worried.
> On 08/01/2016 03:30 PM, Richard Esplin wrote:
> > If nobody cares, why should we waste our time on it?
> > I have submitted lots of bug reports to different open source projects, and I have had many of them age-out and be closed. It's a sad fact that an open bug tracker will receive more issue reports than we have resources to deal with. The software will never be perfect.
> > However, I want to emphasize that many members of our team review every new issue in ALF as they come in. We learn a lot even if we don't get the opportunity to formally act on an individual issue.
> > We appreciate your efforts regardless.
> > Cheers,
> > Richard
> > On Monday, August 1, 2016 1:14:29 PM MDT Andreas Steffan wrote:
> >> On 08/01/2016 12:48 PM, Nick Burch wrote:
> >>> On Mon, 1 Aug 2016, Andreas Steffan wrote:
> >>>> I believe you when you say you'd like to know more about the things we
> >>>> usually apply. Still, it seems you are batch closing issues nobody at
> >>>> Alfresco has ever had a look at.
> >>>> https://issues.alfresco.com/jira/browse/ALF-20508
> >>>> is an example.
> >>>> Almost three years old, still dead easy to reproduce and not even
> >>>> commented.
> >>> If the bug still applies, you ought to be able to hit the re-open
> >>> button, mark it as applying to 5.1 and add a quick comment. I spent
> >>> some of the weekend doing the same for the bugs I've reported while
> >>> still exist and which I still care about!
> >> As I said: Still applies, repro is dead easy and right on the issue.
> >> Nobody cared for almost three years. I don't either today. Nobody will
> >> likely care in the next three years. Pretty amazing if you look closely.
> >>> While it would be lovely if Richard could get all community reported
> >>> bugs reviewed, for now I think it's more useful for us bees if he can
> >>> get a way for us to get our patches reviewed. That way, when we find a
> >>> bug, we can get it fixed much quicker by doing half the work ourselves :)
> >>> Nick
> >> In fact, I think the JIRA issue already is half way. :) Unfortunately, I
> >> found it that "dyi all the way" works best for me. Sad but true.
> >> regards
> >> Andreas
More information about the OOTB-hive