Measurement interval 'on every new commit' does not respect specified branch

[davidkopp]

While looking through the code I stumpled across a potential bug here:

green-metrics-tool/cron/watchlist.py

Line 64 in c39bacb

last_marker_new = utils.get_repo_last_marker(repo_url, 'commits')

If I did not miss anything, it always looks up the latest commit from the default branch of the repository, therefore ignoring the branch specified for the measurement.

[ArneTR]

I agree there could be a bug here.

Do you have an error case / scenario where this lead to a reproducible error?

To my understanding from looking at it now with fresh eyes the errorneous behaviour should rather be the other way around. It should trigger on every commit not only on the specified branch effectively leading to false positives instead of false negatives.

Happy for your feedback

[davidkopp]

Do you have an error case / scenario where this lead to a reproducible error?

No. This case is currently not relevant for me as I want to bring the usage_scenario.yml to the default branch anyway.

To my understanding from looking at it now with fresh eyes the errorneous behaviour should rather be the other way around. It should trigger on every commit not only on the specified branch effectively leading to false positives instead of false negatives.

My understanding is that the current implementation leads to both, false positives and false negatives.
Example (default branch: main, specified branch: energy-tests):

• false positive: a new commit is added to main → measurement is triggered
• false negative: a new commit is added to energy-tests → measurement is not triggered

As a user of the "Submit Software" formular my expection is that the measurement is only triggered if a commit was added to the specified branch. If a commit is added to a different branch (e.g. the default branch of the repository) it should not trigger a measurement.

[ArneTR]

Should be fixed now: d68b768

Since you mentioned that you do not have an error that you are experiencing and this was just from a code deep dive I am closing from my side.

Please re-open if you spot errors