hi all, The 0.11 release push seems to be winding down. I just put up a fairly high priority C++ patch (https://github.com/apache/arrow/pull/2615) containing a bunch of interrelated IPC fixes, careful review requested
Out of the remaining issues, I have questions about: * ARROW-3175 -- upgrading Flatbuffers for Java * ARROW-3141 -- minimum NumPy version in Python wheels * ARROW-1983 -- Writing the _metadata Parquet file from Python The rest of the issues are related to release packaging and verification workflow, or updating the website. It would be great to move the binary artifacts to Bintray but perhaps there is not enough bandwidth for this release, we can always do that for 0.12. Are there any other patches that definitely need to go into 0.11? Any reasons why we could not cut a release candidate this week? Also, can someone volunteer to be the release manager? Thank you, Wes On Mon, Sep 17, 2018 at 1:26 PM Wes McKinney <wesmck...@gmail.com> wrote: > > hi folks, > > It looks like we are closing in on the 0.11 release -- I have spent a > bunch of time gardening the backlog. > > https://issues.apache.org/jira/projects/ARROW/versions/12343066 > > There are quite a few open patches still and housekeeping items > remaining. If you can help complete some of these issues this week, > please take a look and assign yourself to the issue. > > I reckon we need this whole week to get into release-ready shape. Any > reason why we could not have a release vote next week? > > Thanks, > Wes > On Wed, Aug 22, 2018 at 2:31 AM Uwe L. Korn <uw...@xhochy.com> wrote: > > > > Hello, > > > > I will also go over the release and add items. For my personal goal for > > 0.11, I want to have predicate pushdown for Parquet files working. This > > means that we should be able to determine in Python code what the relevant > > RowGroups in a file are as well as filtering a Table given the set of > > predicates. Thus we want to have a pyarrow.parquet.read_table variant that > > takes in a set of predicates in disjunctive normal form and returns a > > (possibly empty) table with only the rows that match the predicates. Please > > be aware when of this when moving issues to 0.12. > > > > Uwe > > > > On Tue, Aug 21, 2018, at 9:01 PM, Wes McKinney wrote: > > > hi all, > > > > > > I know it seems like we just released 0.10, but we are already 75 > > > issues deep in 0.11 and have addressed a number of issues that > > > surfaced with 0.10 > > > > > > https://issues.apache.org/jira/projects/ARROW/versions/12343066 > > > > > > I suggest that we plan to release again the week of September 10. That > > > will be a slow week for me anyway because of the Strata conference in > > > NYC. This gives us enough time to resolve the outstanding packaging > > > issues and to complete the parquet-cpp merge and associated packaging > > > changes needed there. > > > > > > I will start curating the backlog next week to move anything > > > definitely aspirational to 0.12 so we can have a sense of the must-do > > > work for 0.11. Any help with backlog curation would be appreciated. > > > > > > Any other thoughts? In general I think we should be releasing every > > > 4-6 weeks for a while since the pace of progress is quite fast, and > > > many new users are coming into the ecosystem which will benefit from > > > more frequent releases. > > > > > > Thanks, > > > Wes
