Re: [DISCUSS][Julia] How to restart at apache/arrow-julia?

2021-09-20 Thread QP Hou
To expedite the donation, perhaps we could move on with the decoupled version scheme for now to reduce workload and disruption to the existing users. The julia maintainers can always decide to change the versioning scheme later after the donation has been completed. This doesn't seem like a blocker

Re: C++ Parquet thrift_ep No rule to make target install

2021-09-20 Thread Eduardo Ponce
Hi Rares, The reason compilation fails when you set ARROW_PARQUET=ON is because this flag also enables installing Apache Thrift [1] and support for Thrift in CentOS systems is fragile (see THRIFT-2559 [2]). When you disable Parquet, Thrift is not installed as a required dependency. I recommend yo

C++ Parquet thrift_ep No rule to make target install

2021-09-20 Thread Rares Vernica
Hello, I'm compiling the C++ library for Arrow 3.0.0 in CentOS 7. It works fine, but it breaks if I set ARROW_PARQUET=ON. I stops while trying to build thrift_ep > scl enable devtoolset-3 "cmake3 .. \ -DARROW_PARQUET=ON

Re: [DISCUSS][Julia] How to restart at apache/arrow-julia?

2021-09-20 Thread Sutou Kouhei
Hi Jacob, Thanks for confirming this. For major release: As far as I know: We chose this style because we will develop actively in at least a few years. Active development will need API breaking changes. So we release a major version per 3-4 months. Our release process releases all implementat

[C++][DISCUSS] Sequencing in ExecPlan

2021-09-20 Thread Weston Pace
The topic of sequencing to achieve things like deterministic output of head/tail type operations has come up a few times. I went ahead and drew up a proposal for adding these concepts into the exec plan via a batch index. It was just going to be an email but it got a bit long so I put it in a Goo

Re: Proposal to include JS bundle description fix in Arrow 5.0.1

2021-09-20 Thread Dominik Moritz
Okay, that makes sense. Looking forward to making an awesome 6.0 release then. On Sep 20, 2021 at 16:38:40, Neal Richardson wrote: > Unfortunately, I think we've passed the window for doing a 5.0.1. Since > it's not trivial to make a release, and we're doing 6.0.0 in a couple of > weeks (mid-Oc

Re: Proposal to include JS bundle description fix in Arrow 5.0.1

2021-09-20 Thread Neal Richardson
Unfortunately, I think we've passed the window for doing a 5.0.1. Since it's not trivial to make a release, and we're doing 6.0.0 in a couple of weeks (mid-October), in my opinion it makes most sense to focus our efforts on making the next release excellent. Neal On Mon, Sep 20, 2021 at 4:25 PM

Re: Proposal to include JS bundle description fix in Arrow 5.0.1

2021-09-20 Thread Dominik Moritz
Neal, could you make a release? On Aug 27, 2021 at 14:43:10, Dominik Moritz wrote: > Are there any other changes for 5.0.1? It would be great to release a > version with the fixes to the JS bundles since it prevents people from > upgrading from Arrow 3. > > On Jul 30, 2021 at 14:31:59, Neal Ric

Re: [DISCUSS][Rust] Biweekly sync call for arrow/datafusion again?

2021-09-20 Thread Jacques Nadeau
+1 on time variation. Please add me to to the invite. Thanks On Sun, Sep 19, 2021 at 9:49 PM Benson Muite wrote: > New to this. A suggestion may be to consider two of the times, eg. 4:00 > UTC and 16:00 UTC perhaps alternating allowing geographic diversity in > joining convenience. > > On 9/20/

Re: [DISCUSS] Leap seconds/days and day light saving for Duration types

2021-09-20 Thread Rok Mihevc
It looks good to me. Rok On Mon, Sep 20, 2021 at 2:36 PM Antoine Pitrou wrote: > > All, can you please take a look at QP's PR at > https://github.com/apache/arrow/pull/11138 ? > > I don't believe this requires a vote as this clarification is consistent > with the already clarified semantics for

Re: [DISCUSS] Leap seconds/days and day light saving for Duration types

2021-09-20 Thread Antoine Pitrou
All, can you please take a look at QP's PR at https://github.com/apache/arrow/pull/11138 ? I don't believe this requires a vote as this clarification is consistent with the already clarified semantics for Time and Timestamp types. The current PR contents are ready for a merge, and I think