Hi all, I'd like to get a consensus on the 3.0 schedule. We have not yet agreed on the freeze and release dates. There has been some discussion in the irc, and I'll summarize that here as far as I can remember it: I have suggested a freeze date of June 26th. When planning the 2.0 release, we agreed to try the time-based release model with four-month cycles. June 26th is exactly four months after the 2.0 freeze. My suggestion for the release date is "as soon as we have no release blocker bugs and the latest release candidate has been out for long enough". That is, ASAP without any fixed target date. This model can be described as "releases happen every four months, on average". Arun has suggested a release date of September 11th, and freeze a month prior to that (so August 11th would presumably then be it). September 11th is exactly four months after the actual release date of 2.0. This model can be described as "the time between each release is at least four months". David has asked which model would require the least work. I don't remember if Colin has said anything on the matter. My opinion is that deciding the release date is meaningless, because the release will anyway get done when 1) the code is in releasable state and 2) the maintainers have time to do the release routine. A deadline date can maybe help with motivation a bit, but that's it. There's nothing enforcing a strict deadline anyway. I don't mind people setting targets for themselves, of course, but my target will be "ASAP" in any case. As for David's question - I don't see any significant difference in the required work between the two suggested models. In either case all critical bugs have to be fixed - they won't disappear just by waiting a little longer. I imagine rolling the tarballs requires only little work in comparison (I have never done that myself, so I might be wrong). Opinions? -- Tanu
