Hello Suvayu, for arrow-cpp it is definitely possible to cross-compile on the desktop as it using standard CMake for the build. There are a lot of guides available for doing cross compilation with CMake. This may work but I would expect that in some places we're probably not passing all flags through to the third party packages. For pyarrow, we are building all things with CMake but with an indirection. Thus I would expect that cross-compilation will be harder there.
There are currently people preparing a setup where we could build conda packages for ARM, this will be the preferred solution in future. Until then I would suggest that you actually build directly on the PI. This might take a lot of CPU time but will be simplest path for you. Installing just from the pyarrow source packages does not work. Instead you should follow the steps in http://arrow.apache.org/docs/latest/python/development.html using pip. Cheers Uwe On Wed, Jan 9, 2019, at 4:54 AM, Suvayu Ali wrote: > Hi everyone, > > I wanted run a long running data collection process on an RPi. But it > has been proven difficult to install pyarrow with pip as it still needs > to compile. > > Is it possible to I cross-compile it on my desktop? If so, could > someone point me in the right direction? > > Cheers, > > -- > Suvayu > > Open source is the future. It sets us free.
