I'm not convinced (but open to persuasion) that ARROW-8860 alone merits the effort of a patch release. It's unfortunate but has a number of workarounds/alternatives, and it's arguably not a regression but rather a bug in a new feature. Plus, there are (unofficial) nightly Python and R packages available for anyone who must have a fix right away.
Personally, I'd rather see us focus our efforts on getting a rock-solid 1.0 out the door in 5-6 weeks time. Neal On Fri, May 22, 2020 at 1:50 PM Wes McKinney <[email protected]> wrote: > Given some of the biggish projects (eg C++ kernels ARROW-8792) ongoing I'd > be slightly concerned about cutting a release out of master right away > until a little more time has passed, but I can see arguments both ways > > On Fri, May 22, 2020, 3:40 PM Krisztián Szűcs <[email protected]> > wrote: > > > The amount of effort required to create a patch release is comparable > > to a minor release. How about we should create a 0.18 minor release > > instead? > > > > On Fri, May 22, 2020 at 3:54 PM Wes McKinney <[email protected]> > wrote: > > > > > > In light of ARROW-8860 and perhaps some other critical bugs that have > > > been reported, and since our releases have been going more smoothly, > > > what do people think about doing another patch release in a week or > > > two? > > >
