We probably need to support both, conda-forge and CRAN. As a first shot, conda-forge will be much easier to setup as we should have a better build toolchain available there and this could also then be used in the multilanguage scenario demos really well. From my experience, the usage of conda in the R world seems to be very small, so to get a user base, we will need to invest in CRAN.
TL;DR: We should start with conda-forge. Uwe On Thu, Jan 3, 2019, at 10:02 AM, Jonathan Chiang wrote: > > Hi, > > Long term, I believe getting an arrow package onto cran would be most > useful for R users. Building arrow in R on Mac was easier than Linux for > me. I was still having trouble installing after spending a couple hours > or so. > > Typically if you can install.package from cran is most convenient. > Devtools installation from a github repository if it works across > different OSes would suffice. > > Most growing R users are typically using RStudio, so conda may be > inconvenient for R users, because it requires installing IR kernel for > anaconda. > > Whatever installation method or strategy you used for feather was easy > for me to install as a R user. > > Thanks, > Jonathan > > > On Jan 2, 2019, at 11:59 PM, Krisztián Szűcs <szucs.kriszt...@gmail.com> > > wrote: > > > > Perhaps an R conda-forge feedstock? > > I'm not sure how widely conda-forge is used in the R commmunity, > > but it already hosts around a thousand packages[1]. > > > > [1] https://github.com/conda-forge?&q=r- > > > >> On Wed, Jan 2, 2019 at 6:09 PM Wes McKinney <wesmck...@gmail.com> wrote: > >> > >> hi folks, > >> > >> With 0.12 around the corner and significant progress on the R bindings > >> project (sufficient for Spark integration [1]), I am wondering how > >> everyday R users are going to be able to install the software > >> respectively on Linux, macOS, and Windows. Thoughts about the strategy > >> for this? > >> > >> Thanks > >> Wes > >> > >> [1]: https://github.com/apache/arrow/pull/3001 > >>
