Last Saturday Oskar Linden found SVC-8208 and the roll out plans for this week (36) changed. So, we missed a roll out to the main channel, which is old news now. This week things are regrouping and we are in hopes that the bandwidth fixes will make it to the main channel in week 37. I’ll know more later today (Friday), but I won’t get it on the blog until late Saturday.
Oskar was able to reproduce the error (8208) and get the crew started figuring it out and fixing it. Because the code deals with L$ the Lindens considered it serious enough not to proceed with a roll out.
Since Monday was a holiday in the US, Monday’s work and Tuesday’s testing all had to be crammed into Tuesday. We saw this code roll to the Magnum RC channel Wednesday.
Oskar was asked why a fast track channel was being created for something like this 8208. His answer:
We do, but it takes pre-planning. So it goes something like this: [a] bug goes to main channel, we have to find a fix and we don’t see [for until] a few more days that it is a very widespread issue [because it takes time for] stuff [to] bubble up. By then it’s in its own maint-server and was given a priority status. So then it’ll get out the next week.
But then a big issue gets discovered the next weekend and things start piling up. In hindsight pulling it into its own channel would have been a good idea, but we didn’t know at the time how big the issue was. We knew it was annoying and fast-tracked it, but not fast enough for everyone.
[that and the] multiple [failed] RC’s [failing] didn’t help. The best thing would have been for it to have been discovered in RC before main channel promotion.
It’s a complicated dance. People want fixes fixed. We work hard to fix things as fast as possible. [But,] to get to main channel it has to go through RC.
The fix [(bandwidth I think he meant)] was fast tracked. It just didn’t have its own RC. A fix was ready the next week it just hit snags in the RC. I actually called for the team to put the code fix into each RC to give it a higher likelihood of releasing. But, then each RC had showstopper issues. Now, just like last Thursday, it looks like it will go next week.
Tomorrow [(Friday -9/7)] morning is when we discuss promotion. I will suggest making a side branch with just this fix IN CASE of weekend shenanigans. it is extra work in a limited time slot though. Let’s hope the RC gets the promotion next week though. It means other work has to get deprioritized [SIC].
Oskar and Maestro Linden talked about bugs they were finding in the QA process before they made it to the RC channels. One affects objects moving across region boundaries. They lose the root prim. In another case 2 prim objects would lose their child prim. These are being fixed before making it to RC status.
The result is all three RC channels are running the code with new fixes from Magnum… I’m not sure Oskar has that right… it looks like the code from Blue Steel is on all three but it is hard to know which it is. It probably doesn’t matter much as both were very similar.
Discussion turned to the JIRA Change. Oskar thinks it will initially be more work for the QA teams in some ways.
Oskar said, “My biggest concern is that these changes will offer a hindrance to those of you who help me out so much. I need to figure out a way to keep you helping. So given this ‘new JIRA world’ and given the role I play in QA at LL and the roles you play in eager helpful residents we need to figure out a way to continue improving things. If you feel that is impossible in the new system then I need to know.
So, let’s work with the new JIRA for a few weeks and feel it out.
I have no official comment that isn’t in the blog. But I do want to find a way to work with this system, if this is the system we’re going to have.
There are a lot of people who are really helpful and I want those people to continue to be helpful, because it does help. So my job now is to find out how to let the helpful people continue to be helpful in this new setup. So I’m still working with the QA team to sort through how we can continue to be effective with this new system. I say we try it for a few weeks and I’ll see how I can make it work better.”
Oskar made an interesting comment about feature requests.
“I don’t know that feature requests ever really did anything in JIRA. Discussing new features in the forums is probably a better idea. We can see which topics get hotly discussed.
[…]features need to get to the project leads. if project leads read the forums or not, I don’t know.”
Of course this is just Oskar’s take. But, I think it is good information about how to get feature request information to Lindens and a hint at which Lindens.
We’ll know more about the roll outs late Friday.
As to the JIRA, I think it is obvious that the decision was not made at Oskar and Maestro’s organizational level. It looks like they are figuring out how to work with it too.