Discuss/document ci/cd

[tschm]

No description provided.

[tschm]

with @phschiele :-)

[tschm]

How do we feel about pylint? https://pypi.org/project/pylint/

[tschm]

conftest.py is a very useful concept, see https://towardsdatascience.com/testing-best-practices-for-machine-learning-libraries-41b7d0362c95

[phschiele]

with @phschiele :-)

We introduced ruff for cvxpy. It is a powerful linter written in rust.
The first encounter with ruff usually goes like this:

1. You run the tool. It took 0.1s so you question if it actually ran on all files. It did.
2. You change some files locally to violate a rule and see if ruff actually catches it. It did.
3. You are amazed and keep using the tool.

Will create a suggestion.

[tschm]

I think I am doing something stupid in create_kernel.sh. @phschiele any ideas? I also install jupyter-lab as a dev dependency...

[phschiele]

@tschm I'm usually either running jupyter lab directly or open the notebook in VSCode. What is the idea behind calling ipykernel directly?

[tschm]

yes, I call jupyter lab directly. Once in jupyter lab I need the correct kernel. Here I use the kernel I have created before...

[tschm]

branches that have been merged can be deleted? Let's not accumulate our merged branches...

[tschm]

What to do if main is moving fast while you are working on your branch X? You could do a git rebase exercise or you may wanna try merging main into your branch X to get X'. You can then merge X' into main.

[tschm]

branches that have been merged can be deleted? Let's not accumulate our merged branches...

@phschiele I think you could delete the codespaces branch

[phschiele]

What to do if main is moving fast while you are working on your branch X? You could do a git rebase exercise or you may wanna try merging main into your branch X to get X'. You can then merge X' into main.

While rebase can be cleaner in some circumstances by keeping a linear commit history, I usually merge main into the feature branch. This is because once the branch is pushed, others might want to work on it as well, so rebasing could mess up their version of the branch.

[tschm]

Yes, I see. For me feature branches are a very personal thing :-) But rebasing can be a pain as you iterate over all the commits to main you haven't seen yet. You keep correcting the same issue like n.times. Merging main into the feature branch is something I used to consider as some sort of dirty hack but I will adjust my position