tristan wrote:I think I wrote this before but wouldn't it be better to do beta's again, you guys obviously have more than enough work on your plate already. If you like maybe just with a small group of testers? Quality-wise this could improve releases enormously. And it would be an easy way to catch these bugs early on.
We have had a
Beta and a
Release Candidate cycle, and quite a long one at that. And this bug wasn't caught back then. And there were very few participants, although we have had a few quality ones on that group. We have, in addition to that, a public IRC channel for beta testers that we announce at the beginning of each of these cycles, and which is very rarely used, although there are members of the Dev team permanently logged there. So although I'm not discussing the merits of beta testing, we have had very few major benefits from previous ones.
In my personal view, we can, and probably will, keep on doing both cycles on important releases as we have done so far, but doing them for every minor release, doesn't seems productive. I repeat: this is my personal view.
This particular issue with locking has been difficult to get solved because it doesn't affect a majority of the users, has been erratic and difficult to reproduce, and all attempts to solve it have apparently solved it until one user or another keeps finding that it persists. This is the kind of the issues that we want to get caught by a Beta cycle but that wasn't, not this time.
We do have limited human resources on the Dev Team and Beta testing is important, but I don't think there should be one such cycle for every single release, nor that this is the solution to all issues.
Having said that, and as Rolf stated, we are discussing this and trying to find some solution that may be more effective. And we certainly don't want to scare away the ones who have been willing to help, or have been doing so already, quite the opposite. We just need to find the right ways for this to work better.
