02-09-2017 09:36 AM
To complete my answer:
- I think that the way UTF measures the code coverage is good as unit test is not functinal test.
If UTF would measure inside subVIs calls, the results of code coverage would not be representative, and there will be lots of combination t create for test passing, which would always be impacted by changes in code, and would not reflect the non regression.
- Then, if I well understand what you call the "branch coverage", the answer is not code coverage, but the way you create your test and test cases.
You have to create 1 test for 1 branch, and another test for the other branch.
Then, if your two tests are passed, then you have a 100% branch coverage. Is that right?
Olivier L. | Certified LabVIEW Developer
04-03-2018 08:31 AM
Hi!
I found a bug in UTF, which is similar to the mentioned bugs, and the sink of it is may the same as of the bugs before.
In the attached project there is a foo.vi which has two case structure and one Conditional Diagram Disable structure. If I run the created UTF test, the coverage report shows, that there are some uncovered diagram, but if I click them I see, that some of them is covered, and there are no link of some diagrams which are uncovered. If I move the CDD (e.g. into the inner case structure) I get other diagrams as uncovered altough I didn't change the tests.
If I remove the CDD, the list of uncovered diagrams is correct.
Additional info: I found, that the number of uncovered diagrams is the same as the number of the listed diagrams.
Thanks,
Barnabás