A completely unexecuted file shows 14% coverage

[nedbat]

A file with 50 statements and 8 branches, but completely unexecuted, shows as 14% covered. The results code decides that there are no unexecuted branches (because no statements were executed), and comes up with 8 executed branches.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/156

[nedbat]

Original comment by Anonymous

---

I am seeing this same issue. Had upgraded from 3.4 to 3.5.1 and started seeing this (was working at 3.4). Took a look a the diffs in the code files. Found this change to results.py:

40c44,46
< n_missing_branches = sum([len(v) for v in mba.values()])

        n_missing_branches = sum(
            [len(v) for k,v in mba.items() if k not in self.missing]

Seems like this may be the cause ... why would you not count a branch if it was also in the missing statement list? Have to admit I'm a bit over my head trying to analyze the code, but if I revert this change, it seems like I get accurate branch analysis results.

[nedbat]

Fixed in <<changeset cbe93a0d585f (bb)>>. Your branch coverage numbers will be lower... :(