On Wed, Apr 23, 2014 at 7:27 AM, Josh Wills <[email protected]> wrote: > On Tue, Apr 22, 2014 at 12:09 AM, Gabriel Reid <[email protected]>wrote: >> >> Interesting point about potentially doing a 0.9.1 vs 0.10.0. The 0.8.x >> branch seems to be running pretty parallel to the trunk, so it might >> be an interesting idea to keep the point release schedule and >> numbering somewhat aligned. >> >> Any other thoughts on that? >> > > Yeah, not sure. I think we've never really done an 0.x.1 release because so > much was always changing on the API. I know we had a few small-ish changes > this time, but nothing all that major, which made me think that an 0.9.1 > would be okay. This felt like primarily a bug fix release. >
I just took a look through the list of issues for what is currently called 0.10.0, and there are a few that were intentionally left out of the 0.8.x branch to avoid breaking compatibility. The ones I saw were: * CRUNCH-324 - Sample.reservoirSample method name is spelled incorrectly * CRUNCH-357 - Allow AvroMode overrides to be less global * CRUNCH-216 - Transpose arguments in MapsideJoinStrategy.join CRUNCH-324 and CRUNCH-357 are actual changes to the API, so people using those parts of the API won't be able to compile without updating their code -- however, the pieces of the API that are changed are probably not used by that many people. CRUNCH-216 added @Deprecated to one or two methods, with the intention of a gradual API change. This has me (mostly) convinced that we should do a 0.10.0 release instead of a 0.9.1 (although the bulk of the things are still bug fixes and improvements). - Gabriel
