If you like, I can volunteer to kick off a discuss thread when I submit the board report.
On Wed, Aug 15, 2018 at 2:21 PM Michael Miklavcic < michael.miklav...@gmail.com> wrote: > I'm also a fan of the 2-3 month time frame for releases. And I agree it > fits nicely with our board report. That said, I think we should minimally > kick off a DISCUSS at least every 2 months per the recommendations above. > If it's warranted, great. If not, then we bring it up at a stated later > time for re-evaluation. > > Fwiw, some upcoming features post-0.6.0 that I'm seeing which are also > large-ish and will fit nicely into the next cycle (pending completion, of > course): > > 1. NiFi Metron parsers > 2. Profiler enhancements - bootstrapping, etc. > 3. Knox SSO > > > > On Wed, Aug 15, 2018 at 11:10 AM Casey Stella <ceste...@gmail.com> wrote: > > > Strictly selfishly, I'd love for a release to happen quickly enough to > have > > something to announce to the board during the reports. Once every 2 > months > > or when a sufficiently complicated change happens sounds like a sensible > > cadence. > > > > I very much support a "how do we get to 1.0" discussion, maybe as a > > separate thread? > > > > On Wed, Aug 15, 2018 at 11:56 AM zeo...@gmail.com <zeo...@gmail.com> > > wrote: > > > > > I'm a fan of a hybrid time/feature-based cadence. Something like > "When 3 > > > months has passed since our last release, or a sufficiently complicated > > > change has been introduced to master (like merging a FB), a discuss > > thread > > > is started". I'm primarily thinking of what the upgrade path looks > like > > > (more on that in a "how do we get to 1.0" discuss). > > > > > > Jon > > > > > > On Wed, Aug 15, 2018 at 11:02 AM Justin Leet <justinjl...@gmail.com> > > > wrote: > > > > > > > Hi all, > > > > > > > > In concert with the discuss thread on a potential 0.6.0 release, I'd > > also > > > > like start a discussion about our release cadence. We've generally > > been > > > > pretty relaxed around doing releases, and I'm curious what people's > > > > thoughts are on adopting a somewhat more regular schedule. > > > > > > > > Couple questions I think are relevant > > > > 1. Is this something we should work towards and, if we do, how do we > > want > > > > to go about it? > > > > > > > > - "Whenever someone feels like pushing out a discuss thread"? > > > > - "Let's just start a discuss thread every X and if we want to > > release > > > > we release"? > > > > - "let's try to get a release out every X and what's on the bus is > > on > > > > the bus"? > > > > - Something else? > > > > > > > > 2. Assuming we do want to do more regular releases, what's the > > timeframe > > > > we'd like to shoot for? > > > > > > > > Personally, I'd like to just start a discuss thread regularly, with > the > > > > built-in expectation that not every thread should necessarily lead > to a > > > > release. I don't want to be forcing release overhead when there's not > > > > enough to merit a release, but releasing more often than we often do > > now > > > > would provide a lot of values to users. > > > > > > > > In terms of timeframe, I tend to think a 2-3 month cadence for the > > > threads > > > > is reasonable. It's long enough to potentially accrue enough features > > to > > > > merit a release, but short enough that when we pass on a release > we're > > > > probably fine just waiting for another cycle to come around. The > last > > > > release was ~2 months ago and we have a good amount of stuff here, > but > > I > > > > also don't expect two feature branches going in to be the norm. > > > > > > > > I'd expect whatever comes out of this thread to also be relatively > > > > informal. At least right now, I don't feel like we need a rigid > > schedule, > > > > and I'd still like people to feel encouraged to propose a release, > > > > particularly when there are a couple major features or critical > fixes. > > > > Alternatively, I would expect some of these discuss threads to > > conclude, > > > > "We should do a release, but let's wait a couple waits for these > > tickets > > > to > > > > finish up" (e.g. like the Pcap query panel). > > > > > > > > Justin > > > > > > > -- > > > > > > Jon > > > > > >