Description
In March 2019 in RFC-584 we updated python from v3.6 to 3.7. Before the next pipelines release in November I propose we update our baseline python to v3.8 so that we can be prepared for commissioning activities next year. The closer we get to commissioning the less willing we will be to change things and this will give us a solid base to work from.
There are, of course, new features for 3.8 and they are listed at https://docs.python.org/3/whatsnew/3.8.html
These include:
- "walrus" operator
- Improved type annotations
- Improved asyncio (I believe telescope and site are also interested in this)
- Some places that used to return OrderedDict now return a normal dict (because dicts are ordered)
- Speed ups for shutil.copy and shutil.move
Attachments
Issue Links
Activity
+1 Yes please!
!Yes please. If we wait much longer we may as well jump straight to 3.9, which is already in RC.
I think the 3.9 comment is interesting but the real time scale that matters is conda-forge support for 3.9 not the python developers releasing 3.9.
The next pipelines release is November nominally so we could defer to whatever is the newest version in conda-forge in early October.
DM-CCB agreed today that there is no downside to doing the 3.8 transition now.
Supported, we are already having to move to 3.8 in non-stack packages, the sooner we can bump everything up the better.