After this week's 912 update on MOULa, I got to thinking about what should happen about the next update and beyond...
It was notable that following the 905 update in February, CAVCON took a bit of a nose dive for that month. It seems reasonable to infer (but difficult to prove) that was at least partly due to effort expended by Cyan deploying the update. The 906 update was ostensibly "minor" and didn't have any observable effect on CAVCON. It'll be interesting to see what effect, if any, there is on CAVCON this time round, but it's possible that the update occurred too close to the end of the month to appear in the May accounts, so it may be the end of June before we see any effect. I think the verification time was shorter for build 912 than for 905, so hopefully any effect will be reduced.
In any case, what I'm suspecting is that each MOULa update will have some impact on the donation fund, to the extent that aspiring towards a monthly update is probably not something that is sustainable on the current level of donations, so we maybe need to be thinking more in terms of 2-monthly or 3-monthly intervals - that would be 9-weeks or 12-weeks to fit in with the regular re-start cycles. Maybe we don't even need to be that regular, but I thought it might help everyone if there was some kind of longer term view on where updates were likely to occur so we can get PRs in good time to arrange testing and any subsequent fixups/retests prior to Cyan's pickup date and keep everything relatively relaxed. Notionally, and on the preceding basis, I could see either July 30 or August 20 being candidate dates for the next update.
Keep in mind that this is all speculation on my part and we haven't (as far as I'm aware) discussed anything with Cyan regarding when the next update might occur, but it's probably worth discussing what we think would work, then we can see how that fits with Cyan's view and what CAVCON reveals.
Future MOULa updates
Future MOULa updates
Mac_Fife
OpenUru.org wiki wrangler
OpenUru.org wiki wrangler
Re: Future MOULa updates
Another idea that occurs to me reading your analysis (and I thought maybe you were heading here, at first) is that we could also encourage players to bump their donations when updates occur to offset the additional costs. A little community donation drive reminder, perhaps?
Re: Future MOULa updates
Yeah, I'm certainly conscious of the cost side of things, and it certainly wouldn't do any harm to remind people that updates do cost money, despite a level of goodwill from Cyan that I'm sure exists on this (and this weekend's AGM is a good opportunity to make that reminder), but I didn't want to push on that too hard until we saw some better evidence of what the relationship is between updates and CAVCON. So, I concentrated on thinking about "time": I think we have to assume that updates will be aligned with the three-weekly restarts, to get efficient use of Cyan's Support team time, although I initially was thinking in terms of the more intuitive "months" and "quarters" and then converted to restarts which is how I got to the 9-week and 12-week intervals.
Mac_Fife
OpenUru.org wiki wrangler
OpenUru.org wiki wrangler
Re: Future MOULa updates
Just judging from the number of test builds they ran it through this time around, I'm betting that CAVCON will take a bigger dip this month (though we'll still probably only see CAVCON 2 displayed), since they built a total number of 6 builds versus 3 builds previously (excluding 1.906). At least, that's what I suppose the inconsistent jumping build IDs mean.
Lyrositor
Explorer #16601888
To D'ni, or not to D'ni. There is no question.
Explorer #16601888
To D'ni, or not to D'ni. There is no question.
Re: Future MOULa updates
I don't know if it's as simple as that, since the build ID wraps up a lot of things that aren't necessarily "code" (such as exporting the age PRPs) - I know what happened to build 911 but not any of the others. However, I think the real cost comes from "touch time" - the amount of time people actually spend actively working hands-on with the update - and I expect most of that comes during the server update and verification stage where Vicki, Elle, Mike and Mark are all busy for a few hours. Running a build, even if it errors for some reason, probably needs only relatively small amounts of time. Just my opinion.
Mac_Fife
OpenUru.org wiki wrangler
OpenUru.org wiki wrangler
Re: Future MOULa updates
I didn't mean the build time. I meant that multiple builds probably indicate that some errors took place at some point, and that effort had to be spent solving it. So the more builds, the more time taken, the more money spent.
Lyrositor
Explorer #16601888
To D'ni, or not to D'ni. There is no question.
Explorer #16601888
To D'ni, or not to D'ni. There is no question.
Re: Future MOULa updates
It would be nice if we could just ask, instead of constantly guessing even on something trivial like this.
Re: Future MOULa updates
Yes ... I'm all for simplicity and openness as well ...
Re: Future MOULa updates
If we're thinking of a longer term before the next update perhaps we should start throwing out suggestions to Cyan for an update that will have a more visible impact for players than the under the hood stuff we've been working on so far. There has been discussion recently about clothing updates, with new t-shirt designs etc that could easily be implemented. Obviously this raises questions from some about our use of Cyan's max files without have licenses attached to them. However, since the T-Shirts would be appearing on Cyan's shard first it would also serve the added benefit of moving that issue at least one step forward. And there are several in the community (myself included) who have both experience in working with those files and use 3dsMax 7 as their content building environment, and Cyan would need files compatible with max 7.
(The argument with Kaelis and myself in that thread should be disregarded in regards to this issue. It was a personal debate that never should have gotten onto the forums, and I blame myself mostly for that as I should have brought my objections to his post in private to him and not publicly.)
(The argument with Kaelis and myself in that thread should be disregarded in regards to this issue. It was a personal debate that never should have gotten onto the forums, and I blame myself mostly for that as I should have brought my objections to his post in private to him and not publicly.)
-
- Member
- Posts: 317
- Joined: Sat Dec 13, 2008 10:54 am
Re: Future MOULa updates
I wouldn’t try too hard to synchronize our schedule and Cyan’s, as both seem pretty unpredictable. The way it worked out for the 1.912 update was just coincidence.Mac_Fife wrote:I thought it might help everyone if there was some kind of longer term view on where updates were likely to occur so we can get PRs in good time to arrange testing and any subsequent fixups/retests prior to Cyan's pickup date and keep everything relatively relaxed.
I can imagine a usual situation being that we will have several Minkata testing cycles per Cyan update cycle, and possibly not even a whole number. That way, our schedule and Cyan’s are completely independent. Cyan can just pull whenever they feel like it (and the CAVCON fund allows), whatever we have completed at that time, while we're possibly already partway through the next cycle.
Specifically, I was hoping to keep the current Minkata testing cycle a bit shorter than the previous one, so contributors don’t have to wait too long until their changes are in CWE-ou, and others can base their work on them (e.g. it would probably be convenient for many to be able to work in Visual Studio 2010, requiring acceptance of Skoader’s pull request). For the current cycle, I was thinking of including Visual Studio 2010, cursors, Deledrius’ list, clicking-while-sitting. Anything else only if it’s ready sooner than the last of the foregoing, otherwise it goes into the next cycle. Any opinions on that?