fix(hydra): bump Python version to 3.9.1 #2777
Conversation
enaydanov
added
backport/none
| @@ -16,7 +16,7 @@ ENV PIP_NO_CACHE_DIR=1 | |||
| RUN apt-get install -y --no-install-recommends build-essential cmake libssl-dev zlib1g-dev | |||
| ADD requirements-python.txt . | |||
| RUN mkdir /build && \ | |||
| pip3 install -r requirements-python.txt --root=/build --prefix=./ --ignore-installed --no-warn-script-location | |||
| pip3 install -r requirements-python.txt --root=/build --prefix=./ --ignore-installed --no-warn-script-location --use-feature=2020-resolver | |||
There was a problem hiding this comment.
I've help tested this feature in europython2020 sprints... it's helpful for our complex requirements.txt
requirements-python.txt
Outdated
| selenium==3.141.0 | ||
| mysql-connector-python==8.0.19 | ||
| docker==4.2.0 | ||
| python-jenkins==1.4.0 | ||
| ssh2-python==0.18.0.post1 | ||
| ssh2-python==0.22.0 |
There was a problem hiding this comment.
when they did released this one ?
There was a problem hiding this comment.
They released 0.20.0, 0.21.0 and 0.22.0 in last three weeks. But major change was 0.19.0 in mid-August.
|
@enaydanov BTW, seem like we'll need to address some pylint related issues (my guess bug in pylint): |
No, there is no rush. This PR is just a kind of a placeholder for this work. And a research work. Eventually, we'll switch to Python 3.9. Bentsi and I decided to have a presentation about new features in Python 3.8 (because we missed this) and 3.9. Maybe next week. After that we can move forward and switch to the new version. Personally, I think it should be like "Why we need to keep old version?" instead of "Why we need to switch to the new one?" Software evolving rapidly and, "my dear, here we must run as fast as we can, just to stay in place. And if you wish to go anywhere you must run twice as fast as that." :)
Yes, you're right: pylint-dev/pylint#3882 And it's a showstopper for us. It's better to wait for them to fix it. |
|
And there is another one pylint issue: pylint-dev/pylint#3876 |
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
from
ee6435b
to
b9c2d79
Compare
|
I've played a liitle bit with pylint and added a pretty simple workaround for linting issues with |
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
from
b9c2d79
to
f5929f3
Compare
|
@enaydanov I think we should take it for a few spins on GCE and AWS first. (even docker) |
| # * https://github.com/PyCQA/pylint/issues/3876 | ||
| # Once these issues will be fixed we can update pylint and remove this plugin. | ||
|
|
||
| from astroid import MANAGER, FunctionDef, inference_tip, extract_node |
There was a problem hiding this comment.
from my experience astroid tend to break too often, I would rather wait for a fixed version. (or dive in, and fix it at the source of it)
the only reason I would consider pylint plugins, is for writing our own checks, but I don't have any idea of what we could check, we should consider it regardless of moving to python3.9
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
2 times, most recently
from
72d8c0b
to
42a125e
Compare
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
from
42a125e
to
ed47c73
Compare
enaydanov
changed the title
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
2 times, most recently
from
0243846
to
1d9351b
Compare
enaydanov
force-pushed
the
hydra_python_3.9.0
branch
from
1d9351b
to
9eb67f9
Compare
https://www.python.org/downloads/release/python-390/
What's New in Python 3.9.0 (sorted by probability of facing usage of, in my opinion):
In media:
Open Python 3.9 pylint issues:
PR pre-checks (self review)
backportlabelsNew configuration option are added and documented (insdcm/sct_config.py)I have added tests to cover my changes (Infrastructure only - underunit-test/folder)