The 'stable-1' pointer was a hedge at the time we decided to make the 'stable' pointer point at a branch-2.x for the first time. The experience has been good, so there isn't a need to hedge any longer, in my opinion.
We should delete the stable-1 pointer. On Mon, Mar 2, 2020 at 6:41 AM Sean Busbey <[email protected]> wrote: > I don't remember who I was chatting with, but the stable-1 pointer > came up and it reminded me that I don't care for it. :) > > As a community we use the "stable" pointer as a way to guide > downstream folks who don't want to be actively engaged in HBase > internals. It's supposed to be a guidepost that says "this is our best > bet on you having a low-pain experience." Right now our stance is > that's a HBase 2 release. > > What purpose does "stable-1" serve? I previously was an advocate for > it as a way to say "hey if you have to stay on HBase 1 then use this > one." But as our collective effort on HBase 1 releases has waned I > think the answer to that increasingly becomes "use the latest HBase 1" > because we effectively can't sustain more than a single branches-1 > based release line. > > AFAIK no one is prepared to do that kind of extensive vetting of a > branch-1 based release that would e.g. justify having folks stick to > 1.4.z releases instead of updating to 1.6.0 when it comes out. > > I'm happy for various HBase 1.y lines to keep going so long as there > are RMs willing to step up. I still think we should have monthly 1.4.z > releases through June. But if we get into a regular cadence of > releases off of branch-1 I'd rather we not provide folks with a mixed > message about wether or not they'd be better of going to those newer > release lines. > > What do folks think? Should we just delete the stable-1 pointer? > -- Best regards, Andrew Words like orphans lost among the crosstalk, meaning torn from truth's decrepit hands - A23, Crosstalk
