What's the timeline for 3.8 support? Kedro is currently marked as <3.8, but I assume few problems exist to allow 3.8?
New version of Python.
## Possible Implementation
-
## Possible Alternatives
-
@Pet3ris Thank you for opening the feature request! We will discuss it and the timeline internally :)
@Pet3ris We had a stab at it, and encountered some compatibility issues downstream. Specifically, numpy C-modules are not being installed properly on Python 3.8 on conda. We will keep track of it and get back you when we have any progress.
Update: conda & numpy seem to be fine. Currently blocked by https://github.com/apache/arrow/pull/5685
Have Arrow started building 3.8 wheels? See here: https://github.com/apache/arrow/commit/6c6754aa5fb840466dfd618ca5b6ae7ca72ebbfa. Does this mean 3.8 is good to go?
FYI @921kiyo @DmitriiDeriabinQB @lorenabalan
The latest pyarrow version 0.15.1, unfortunately, does not come with Python 3.8 wheels, so running pip install pyarrow fails on 3.8 at the moment: "Best guess would be end of January but it may be sometime in February"
Hey guys, latest commit on develop aims to enable 3.8 compatibility https://github.com/quantumblacklabs/kedro/commit/fe489536398408761274f6c1e3d820649b7c0b5e - feel free to have a play and let us know if you find any issues. Please note that pyspark is a known issue.
Thank you for the feature request! I am now closing this as Python 3.8 compatibility was enabled starting with version 0.16.0 released today.
Most helpful comment
Hey guys, latest commit on develop aims to enable 3.8 compatibility https://github.com/quantumblacklabs/kedro/commit/fe489536398408761274f6c1e3d820649b7c0b5e - feel free to have a play and let us know if you find any issues. Please note that
pysparkis a known issue.