2. Yes. We shouldn’t leave a regression out in the wild. Review/merge and immediate VOTE.
If possible, we should also mark 3.4.5 as a deprecated release on maven central and other repositories. On Thu, Feb 13, 2020 at 1:01 AM Jorge Bay Gondra <[email protected]> wrote: > Thanks for starting the discussion. It's a good thing that it was > identified so quickly, I think we should tackle this asap to affect the > least amount of users and providers. > > 1. yes. > 2. review/merge and then immediate VOTE. I think we shouldn't wait for > other potential issues. > > On Wed, Feb 12, 2020 at 10:25 PM Stephen Mallette <[email protected]> > wrote: > > > To my recollection we haven't released with a bug that made me think > about > > immediately re-releasing a patch since 2011, but it looks like 3.4.5 has > > something not so nice in it. It was brought to light on gremlin-users > today > > and I created this JIRA for it: > > > > https://issues.apache.org/jira/browse/TINKERPOP-2338 > > > > While the bug has a workaround it's not nice and existing Gremlin > matching > > that pattern will immediately stop behaving properly on upgrade. Given > that > > the bug is related to graph mutations I find that a bit high on the > > worrisome scale. > > > > I suppose the good news is that we learned about the problem quickly > before > > a ton of new code went to 3.4-dev and there is a fix that is surgical and > > touches very little code. So, we could reasonably get this change through > > review, merge it to 3.4-dev, create 3.4.6 artifacts and prepare for > VOTE. I > > don't see the need for a week long code freeze - we just skip to release > > VOTE. > > > > So, I suppose the points for discussion here are: > > > > 1. Is the bug bad enough to warrant an immediate release of 3.4.6 (note > > 3.3.11 is unaffected)? > > 2. If so, are we good with a quick review/merge and an immediate build > of a > > 3.4.6 artifact for VOTE or should we perhaps take a bit more time to see > if > > other bad news comes in? > > > -- Divij Vaidya
