Thanks. I looked at it. I think I know how it works and the amount of plumbing should be fairly minimal.
Using reticulate from rpy2 is not an option because of a long-standing issue (segfault when reticulate trying to initialize Python). I am unclear about how to bridge cffi and Cython though. Is there an example of using the C-code in pyarrow from a Python C-extension? (the same C structures and functions already made accessible to Cython). Le sam. 2 janv. 2021 à 19:20, Wes McKinney <[email protected]> a écrit : > We can go R to/from Python from an R perspective with reticulate > > https://arrow.apache.org/docs/r/articles/python.html > > There are unit tests attesting to this working: > > https://github.com/apache/arrow/blob/master/r/tests/testthat/test-python.R > > However, I don't know of anyone trying to get interop from a Python > perspective with rpy2; there may be a small amount of plumbing needed > to get it working, others may know more. > > On Sat, Jan 2, 2021 at 5:20 PM Laurent Gautier <[email protected]> wrote: > > > > Hi, > > > > I am looking at sharing a pointer between Python and R. For example > create an Arrow object with Python, perform initial filtering, and then > pass a shared pointer to R through rpy2 (meaning that an R6 object is > created from this pointer and R package arrow). > > > > I found in the source a file that suggest thats Python-to-R is either > planned or may be even already functional: > https://github.com/apache/arrow/blob/master/r/src/py-to-r.cpp > > However, I did not find documentation about it. > > > > Would anyone here know more about this? > > > > Best, > > > > Laurent > > > > >
