None of them is perfect, but here are the main differences:
Drawback: the Git history will be polluted by automated git commit -m "Updated coverage.svg";
Advantage: completely automated.
Drawback: need to be run manually on one's local machine and the files coverage.xml and coverage-badge.svg need to be commited;
Advantage: clicking on the badge should show an HTML report where one can visualize the testing summary and see which lines of code is tested and which isn't, like in https://github.com/openai/gym/issues/2734#issuecomment-1094573459 (the folder where the web report files are needs also to be commited).
Would anyone have any preference between those two solutions?
Personally, I would really prefer if the coverage could be done on github side with the CI and if that is no possible then I would be tempted not to use it until then.
As https://github.com/openai/gym/pull/2789 is becoming a dead-end because of Github permissions issues, I started to look into other solutions from https://github.com/openai/gym/issues/2734, like the ones below:
None of them is perfect, but here are the main differences:
git commit -m "Updated coverage.svg"
; Advantage: completely automated.coverage.xml
andcoverage-badge.svg
need to be commited; Advantage: clicking on the badge should show an HTML report where one can visualize the testing summary and see which lines of code is tested and which isn't, like in https://github.com/openai/gym/issues/2734#issuecomment-1094573459 (the folder where the web report files are needs also to be commited).Would anyone have any preference between those two solutions?