Open Carreau opened 4 years ago
(also we should really test on 3.8 and 3.9).
Hi @Carreau - thanks for raising this issue.
Personally, I think its too soon to drop 3.6 and would rather we adhere to the general python-support timeframes: https://devguide.python.org/#status-of-python-branches
The Notebook 6.1.0 is in progress with rc1 released last week. I think we could probably entertain dropping 3.5 from 6.1.0 given its end-of-life in a few months, but dropping 3.6 seems premature.
I agree that we should add formal support for 3.8 and look at enabling testing for 3.9. (I'll look into this.)
What do others think wrt to 3.5 and 3.6?
No objections from keeping especially if close to release, Just getting the conversation started. I haven't contributed to notebook is forever so will trust you.
I think that 3.5 is relatively old though already, and I am not sure there is much advantage of keeping it. 3.6 I'm on the fence and can understand if some project want to be more conservative than others.
NEP 29 (numpy/numpy#14086) suggest dropping Python 3.6 support soon. Some of Jupyter dev support it and have signed the NEP.
Dask is considering it, IPython will definitely do it.
Should we consider that consider that as well for the notebook (/notebook server) ? That would mean dropping support for 3.5 and 3.6 in next version.