skip_covered changes reported total percentage

[nedbat]

Originally reported by Charlie Denton (Bitbucket: meshy, GitHub: meshy)

---

A codebase I manage has noticed that since upgrading to 4.0.0 there is a drop in reported coverage (from 99% to 97%).

This value goes back to 99% when I remove the skip_covered setting from .coveragerc.

Unfortunately, I can't yet provide an example of this on an open source project.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/423

[nedbat]

Original comment by Florian Bruhin (Bitbucket: The-Compiler, GitHub: The-Compiler)

---

There's some discussion about this in #43 - it seems this is intentional and the total only shows the visible rows.

[nedbat]

Pull request 43 (https://bitbucket.org/ned/coveragepy/pull-requests/43/add-skip-covered) has some discussion, but we were undecided, and expected feedback. Seems like we are getting feedback :)

Looking at it now, it does seem wrong that the total would exclude the skipped files. --fail-under is also affected, which seems wrong. The discussion in the pull request contrasts this feature with the HTML filtering, but HTML filtering is interactive and ad-hoc, for exploring the results. --skip-covered is a way to remove noise from the output.

[nedbat]

Fixed in 0d66850800e2 (bb).