Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Drop Python 3.5? #1193

Closed
arporter opened this issue Apr 7, 2021 · 16 comments
Closed

Drop Python 3.5? #1193

arporter opened this issue Apr 7, 2021 · 16 comments
Assignees
Labels

Comments

@arporter
Copy link
Member

arporter commented Apr 7, 2021

The notebook tests are failing for Python 3.5. Possibly this is because Jupyter has dropped support for Python 3.5 (jupyter/notebook#5962). Apparently 3.5 was EOL last September.

As discussed at jupyter/notebook#5961 (comment)), we could choose to pin the version of Jupyter we use but I think a better option is to drop 3.5.

@arporter
Copy link
Member Author

arporter commented Apr 7, 2021

For now, I've made the success of the Python 3.5 tests an optional requirement before merging a branch to master.

@hiker
Copy link
Collaborator

hiker commented Apr 7, 2021

I'm ok with dropping 3.5 (Intel OneAPI comes with python 3.7.9, and I am using 3.8).

@sergisiso
Copy link
Collaborator

Even if Python 3.5 is not marked as required now, when it fails it cancels the other tests and make the whole CI not pass.

@rupertford
Copy link
Collaborator

If we are going to do this then it should be made clear in the documentation that it does not work. Also, what version of Python3 do the Met Office use and what version is in the LTS versions of linux? If 3.5 is still a standard version anywhere then I don't think we should drop it. I also thought that Sergi found that this problem had gone away (and I've not noticed it yet) - obviously not?

@TeranIvy
Copy link
Collaborator

TeranIvy commented Apr 7, 2021

We're using 3.6.8 in LFRic, so from our point of view it makes sense to replace 3.5 testing with 3.6.

@arporter
Copy link
Member Author

arporter commented Apr 8, 2021

Apparently 3.5 is shipped with Ubuntu 16.0.4 which (I've just learned) reaches EOL this month! Therefore, it may be premature to drop support for it. Strictly speaking, we just wouldn't be running the test suite for it any more - we could still support it if a user appeared who needed it.

arporter added a commit that referenced this issue Apr 8, 2021
@arporter
Copy link
Member Author

arporter commented Apr 8, 2021

Suddenly this problem has gone away again! I've delved in the logs for the tests and there are differences in the dependencies that are being installed:

$ diff working_list.txt broken_list.txt 
< decorator-5.0.6-py3-none-any.whl
> decorator-5.0.5-py3-none-any.whl
< jupyter_client-6.1.12-py3-none-any.whl
> jupyter_client-6.1.13-py3-none-any.whl
> nest_asyncio-1.5.1-py3-none-any.whl

@arporter
Copy link
Member Author

arporter commented Apr 8, 2021

Anyway, we can close this for the time being.

@arporter arporter closed this as completed Apr 8, 2021
@arporter arporter reopened this Sep 30, 2021
@arporter
Copy link
Member Author

@LonelyCat124 is getting failures again with 3.5 which is now a year past its EOL. 3.6 reaches EOL this year too. I propose we drop 3.5...

@arporter
Copy link
Member Author

and @LonelyCat124 also points out that 3.10 is due in 5 days!

@LonelyCat124
Copy link
Collaborator

Ubunbu 18.04 LTS end their "Hardware and maintenance updates" 6/2022, and ships with Python 3.6 as their default Python version, so I've bumped 3.5 up to 3.6 in #1415 as discussed with Andy

@arporter
Copy link
Member Author

Flake8 is now failing with 3.5 and I can see no reason (given the above discussion) not to switch from 3.5 to 3.6 so I'll open a PR.

@sergisiso
Copy link
Collaborator

We could also consider bumping 3.8 to 3.10 in GHA, so we check with the oldest and latest stable versions.

@arporter
Copy link
Member Author

Yes, I'd be happy with that. Any objections from @TeranIvy, @rupertford or @hiker ?

arporter added a commit that referenced this issue Nov 25, 2021
arporter added a commit that referenced this issue Nov 25, 2021
arporter added a commit that referenced this issue Nov 25, 2021
This reverts commit 954801d.
arporter added a commit that referenced this issue Nov 25, 2021
@hiker
Copy link
Collaborator

hiker commented Nov 26, 2021

OK with me.

@arporter
Copy link
Member Author

Reopening as I've hit trouble with the fact that we pin the version of jsonschema in setup.py and there's a TODO there for this issue (naughty us missing that).

@arporter arporter reopened this Jan 17, 2024
arporter added a commit that referenced this issue Jan 17, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

6 participants