thoughts on sheriffing
I sheriffed yesterday, on the last day before the code freeze for Firefox 3, beta 4. Sheriffing is an unusual role within the Mozilla community in that it isn't distributed meritocratically. It's a shared responsibility of the development team, and you can be assigned to sheriff without a clue how to do it.
(That said, it's not clear how one gets on or off the roster. We'd benefit from a clear and simple policy here, like making it be the set of people who have checked in to Gecko and Firefox modules recently, or some other set that represents the most likely current stakeholders in the health of the tree.)
And it's a pressure cooker, with lots of hard decisions to make, plenty of chances to screw things up, and a bevy of developers clamoring for a more or less open tree depending on their individual circumstances and biases.
Nevertheless, it's great training in making hard decisions under pressure (with the safety net of revision control, unlike in the real world, where you can't uncut the red wire), and there are plenty of people helping out with advice and assistance with the chores of sheriffing.
Reed Loden and David Baron, in particular, were a huge help to me yesterday in tracking down regressions, backing out patches, filing bugs, and the like. And David ultimately took on the last leg of sheriffing when we extended the freeze a few hours to make up for a late tree closure (the third that day). Thanks guys!
(Thanks too to Matthew Zeier for jumping in at midnight to kick a box that was preventing the tree from reopening!)
Ultimately, despite the difficulties associated with an all "volunteer" force, I don't think I would professionalize the sheriff role. The tree is our shared treasure, and sheriffing is a great eye opener to the project-wide costs of individual mistakes and the value of good tree etiquette, helping us develop into better stewards of the source. I think it's worth the pain, but for a different perspective, see dougt's thoughts.
Update: dougt's blog post has since been removed, but he's expressed his current thoughts on the subject in a comment on this post.
(That said, it's not clear how one gets on or off the roster. We'd benefit from a clear and simple policy here, like making it be the set of people who have checked in to Gecko and Firefox modules recently, or some other set that represents the most likely current stakeholders in the health of the tree.)
And it's a pressure cooker, with lots of hard decisions to make, plenty of chances to screw things up, and a bevy of developers clamoring for a more or less open tree depending on their individual circumstances and biases.
Nevertheless, it's great training in making hard decisions under pressure (with the safety net of revision control, unlike in the real world, where you can't uncut the red wire), and there are plenty of people helping out with advice and assistance with the chores of sheriffing.
Reed Loden and David Baron, in particular, were a huge help to me yesterday in tracking down regressions, backing out patches, filing bugs, and the like. And David ultimately took on the last leg of sheriffing when we extended the freeze a few hours to make up for a late tree closure (the third that day). Thanks guys!
(Thanks too to Matthew Zeier for jumping in at midnight to kick a box that was preventing the tree from reopening!)
Ultimately, despite the difficulties associated with an all "volunteer" force, I don't think I would professionalize the sheriff role. The tree is our shared treasure, and sheriffing is a great eye opener to the project-wide costs of individual mistakes and the value of good tree etiquette, helping us develop into better stewards of the source. I think it's worth the pain, but for a different perspective, see dougt's thoughts.
Update: dougt's blog post has since been removed, but he's expressed his current thoughts on the subject in a comment on this post.
posted by Myk at 11:51 AM
4 comments
links to this post
