Too many lords, not enough stewards
Too many lords, not enough stewards
Posted Feb 2, 2018 17:55 UTC (Fri) by seanpaul (guest, #70610)In reply to: Too many lords, not enough stewards by error27
Parent article: Too many lords, not enough stewards
As a member of a maintainer group myself, I can tell you definitively that I am as much of a jerk as I was before. That said, if I am treating someone unfairly, I fully expect my counterparts to let me know that. I promise to do the same if another maintainer is having a bad day.
Most people are not jerks most of the time, instituting checks and balances really does help.
> Me: This patch that fixes a NULL deref
> Dev: F you! Why did you mark my patch with the fixes tag!
This would not fly on dri-devel, group maintainership or not.
(Log in to post comments)
Too many lords, not enough stewards
Posted Feb 3, 2018 10:05 UTC (Sat) by jani (subscriber, #74547) [Link]
> I am as much of a jerk as I was before. That said, if I am treating someone
> unfairly, I fully expect my counterparts to let me know that. I promise to
> do the same if another maintainer is having a bad day.
>
> Most people are not jerks most of the time, instituting checks and balances
> really does help.
Agreed. Having your maintainer peer tell you "hey, that's not cool" is hard to shrug off.
Too many lords, not enough stewards
Posted Feb 4, 2018 3:43 UTC (Sun) by error27 (subscriber, #8346) [Link]
In retrospect here is probably where that email thread went wrong. It was the kind of bug where you forget to set an error code and then do return ERR_PTR(rc). That is a NULL return essentially. The callers don't check for NULL and crash.
Apparently it was documented somewhere that callers were supposed to check, but originally the code only returned error pointers and one caller didn't check for NULL. His bug of accidentally returning NULL should have been harmless. A later caller was added that also didn't check for NULL.
I gave the fixes tag to his patch which introduced the runtime bug. He wanted me to give it to the later caller that was added. That seems obviously wrong, but possibly giving it to the other earlier patch which didn't check for NULL would have been correct.
If a third party had got involved we could have figured what was going on without the annoyed email thread. Or it would help to just have someone else acknowledge that the code is subtle and it's not always obvious where fixes tags should go so let's go easy on each other.
In the end we figured out the technical bits and he asked me to redo the patch, and I told him what, after you were so rude to me? Do it your own self.
These kinds of conflicts are a normal part of life and I don't have any lasting ill will. To be honest, I don't remember the other developer's name. It wasn't someone I deal with a lot.
But I do like the idea of bringing in a calm person when it feels like someone is starting to get annoyed instead of waiting until it's too late.