The conda packages from the default channel are indeed managed by Anaconda (Continuum Analytics is their old name). For arrow-cpp, the reciple is here: https://github.com/AnacondaRecipes/arrow-cpp-feedstock
And indeed, looking at the windows build script, LZ4 is not enabled: https://github.com/AnacondaRecipes/arrow-cpp-feedstock/blob/master/recipe/bld.bat You can open an issue there to try to reach out to them. Joris On Fri, 11 Jun 2021 at 10:38, Joris Peeters <joris.mg.peet...@gmail.com> wrote: > > In the matrix of pyarrow artifacts for [conda, conda-forge] x [linux, > windows], it seems all of them support LZ4, except (conda, windows). Is there > a known reason for this? > > As I understand it, the conda Arrow builds are managed externally (by > Continuum?), so perhaps should get in touch with them - as might not have > `-DARROW_WITH_LZ4=ON`? > > There errors at run-time look like, > > ... > pa.CompressedInputStream( > File "pyarrow\io.pxi", line 1208, in > pyarrow.lib.CompressedInputStream.__init__ > File "pyarrow\io.pxi", line 1579, in pyarrow.lib.Codec.__init__ > File "pyarrow\error.pxi", line 141, in > pyarrow.lib.pyarrow_internal_check_status > File "pyarrow\error.pxi", line 118, in pyarrow.lib.check_status > pyarrow.lib.ArrowNotImplementedError: Support for codec 'lz4' not built > > Cheers, > -Joris. >