Conversation
52856dd to
105df42
Compare
Codecov Report
@@ Coverage Diff @@
## master #38 +/- ##
==========================================
+ Coverage 77.76% 79.20% +1.44%
==========================================
Files 5 6 +1
Lines 697 755 +58
==========================================
+ Hits 542 598 +56
- Misses 155 157 +2
Continue to review full report at Codecov.
|
33dc072 to
d079163
Compare
709726b to
121e502
Compare
|
Major changes:
Open questions:
|
0f5dd5e to
3208bac
Compare
|
TODO: clean up We also have to make sure that these things cannot be mutated. |
607e79e to
5d9a2c4
Compare
|
I think this is ready for review @ctrueden. I added a flaking job to Github Actions; this enforces code style. You can check the code style locally using |
The github actions script now is much more similar to PyImageJ's.
It is not used
This makes it easier to read
We were already depending on it for code coverage, so let's just make the full switch
We instead rely on setuptools_scm to manage the files added to the
sdist.
The discussion on whether test and documentation files belong in the
sdist is an ongoing one (see
https://discuss.python.org/t/should-sdists-include-docs-and-tests/14578?page=5
), and it's my impression that the participants in the dicussion want to
leave this up to the tool building the sdist. So we will do that!
Since we cache by default, this decorator was intended to establish constants. Of course, in the wild west that is Python, someone could just "overwrite" our constants, but it helps to declare our intentions with the functions being decorated
Function calls as default arguments are really misleading. Default arguments are evaulated at function definition time, NOT when the function is called. While in this particular case, we are accessing a global variable, leading to a case where no harm is done, we should avoid this generally speaking.
This is a common convention in Python to indicate that variables are deliberately discarded. See https://stackoverflow.com/a/5893946
|
Yay! Thanks, @gselzer! |
This PR is designed to incorporate the lessons learned from improving the build process of pyimagej and imglyb.