"Gardening" old bugreports

Nate Graham nate at kde.org
Thu Jan 19 16:49:38 GMT 2023


Hello folks,

I did approve this initiative and I do think there's value to it, but of 
course we can definitely have this discussion in the open and tweak its 
parameters, or end it if we think it's destroying more value than it's 
creating.

I totally agree with Nicolas that in an ideal world, all bug triaging 
would be done on a bug-by-bug-basis, and this is what I personally do 
with products I'm familiar with and have some technical knowledge about. 
But I also can't blame anyone else who doesn't do this, because manual 
bug triaging is a tedious, laborious, time-consuming, energy-draining 
task. It's hard enough for new bugs, and for old bugs which were 
poorly-handled in the past, lack key information, or are full of angry 
users, it can be even more unpleasant.

In my experience mentoring volunteer bug triagers over the past 5 years, 
few stick around for manual bug triage longer than about 3 months. It's 
not fun and it feels like work. It's hard enough even to get maintainers 
to do it consistently, for the same completely justifiable reasons. I 
have tremendous respect for those who grit their teeth and do it anyway.

But it's currently not enough to handle the volume of bug reports we get 
daily, or reduce the backlog of old un-handled bugs. This causes 
Bugzilla's signal-to-noise ratio to worsen over time. The automatic bug 
triage initiative emerged to try to address that. If our consensus ends 
up being to terminate it, I'd like it to be because more folks are 
stepping up to make manual bug triage a part of their daily routine--and 
not just for new bugs, but for old ones too. And on a consistent basis, 
not just once or twice. The more human labor we get consistently working 
on the problem, the less drive there will be for machines to do it.

Here's our documentation on how to do it: 
https://community.kde.org/index.php?title=Guidelines_and_HOWTOs/Bug_triaging

If that doesn't happen, I think we need to consider other options for 
bug management, such as doing it automatically or hiring more people to 
do it manually.

Let me know your thoughts!


Nate


More information about the kde-devel mailing list