Thanks for commenting, all. I'll open a JIRA/PR to remove support next week.
On Tue, Aug 10, 2021, 09:34 Wes McKinney <wesmck...@gmail.com> wrote: > +1 for dropping it also. > > On Mon, Aug 9, 2021 at 7:03 PM Keith Kraus <keith.j.kr...@gmail.com> > wrote: > > > > +1 as well. Is there any build platforms that we're currently supporting > > that still use vs2015? > > > > Conda-forge did its migration ~1.5 years ago: > > https://github.com/conda-forge/conda-forge-pinning-feedstock/pull/501. > > > > -Keith > > > > On Mon, Aug 9, 2021 at 12:01 PM Antoine Pitrou <anto...@python.org> > wrote: > > > > > > > > +1 for requiring a more recent MSVC version. > > > > > > Regards > > > > > > Antoine. > > > > > > > > > Le 09/08/2021 à 17:38, Benjamin Kietzman a écrit : > > > > MSVC 19.0 is buggy enough that I for one have spent multiple days > > > > reworking code that is fine for all other compilers we test against. > > > > Most recently in the context of > > > https://github.com/apache/arrow/pull/10793 > > > > (ARROW-13482) I found that for some types T, > > > > `std::is_convertible<T, T>::value` will be false. This necessitated > the > > > > following > > > > (very hacky) workaround: > > > > > > > > > > > > https://github.com/apache/arrow/pull/10793/commits/c44be29686af6fab2132097aa3cbd430d6ac71fe > > > > > > > > (Side note: if anybody has a better solution than that specific > > > hack, > > > > please don't hesitate to comment on the PR.) > > > > > > > > Would it be allowable for us to drop support for this compiler? IIUC > > > > Microsoft is no longer accepting feedback/bug reports for VS2017, let > > > > alone VS2015. Are there any users who depend on libarrow building > > > > with that compiler? > > > > > > > >
