Lambda expression confuses branch measurement

[nedbat]

A line like this:

    fn = make_sortable or (lambda x: x) 

is marked as being a branch with a missing jump to "exit", maybe only if the lambda isn't used.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/90

[nedbat]

Original comment by Carl Meyer (Bitbucket: carljm, GitHub: carljm)

---

I can verify that, at least in my case, the missing jump to exit disappeared when I ensured the lambda was used during the test run. Seems to me this could be considered useful coverage information, not a bug -- if anything, the branch-coverage report should special-case this to display a more informative label on the line?

[nedbat]

A particularly confusing form of this bug:

OPERATIONS = [
    operator.add,
    operator.sub,
    lambda x: 2*x,
    operator.mul,
]

The "OPERATIONS = " line will have a missed branch to "exit" if the lambda expression is never executed.

[nedbat]

Bug #460 is related to this.

[nedbat]

Perhaps the best we can do here is to report it as "Missed lambda" instead of "missed branch to exit."

[nedbat]

It's not as good as I would like, but as of now, the HTML shows "3 ↛ exit", and the hover text reads "line 6 didn't finish the lambda on line 6".

I think this provides enough information to the reader. Getting more precise would require collecting data differently, to distinguish the line 6 execution during building the list, from line 6 execution during running the lambda. We can distinguish them based on the co_firstlineno of the frame, but we don't have a way to record that data as we go.

[nedbat]

Original comment by Daniel Hahler (Bitbucket: blueyed, GitHub: blueyed)

---

Could this be reflected in the xml report, too?

[sigma67]

Has this been fixed? I'm still seeing issues.

conversion_lookup: Dict[Type[Union[int, bool, float, str, Path]], Callable[[Any], str]] = {
        int: str,
        bool: str,
        float: str,
        str: str,
        Path: lambda x: str(x.as_posix()),
    }

coverage 7.2.7

[nedbat]

@sigma67 since this ticket is so old, it would be better to open a new issue with complete details of what the problem is and how to reproduce it.

[sigma67]

The issue was actually on my end, the lambda function in question was in fact not covered. Sorry for bothering you about it.