Miss-match between testrunner's coverage console output & generated HTML report

[stoyandimov]

I'm running report generator against a opencover report.

The console output:

+------------------------------+--------+--------+--------+
| Module                       | Line   | Branch | Method |
+------------------------------+--------+--------+--------+
| Unic.Services.AudioConverter | 59.25% | 64.7%  | 51.45% |
+------------------------------+--------+--------+--------+

+---------+--------+--------+--------+
|         | Line   | Branch | Method |
+---------+--------+--------+--------+
| Total   | 59.25% | 64.7%  | 51.45% |
+---------+--------+--------+--------+
| Average | 59.25% | 64.7%  | 51.45% |
+---------+--------+--------+--------+

Expected behavior

ReportGenerator HTML report showing Unic.Services.AudioConverter:

• Line coverage: 59.2* %
• Branch coverage: 64.7* %

Actual behavior

ReportGenerator HTML report shows Unic.Services.AudioConverter:

• Line coverage: 59.2 % (as expected)
• Branch coverage: 68.2 % (not as expected)

Screenshots

Here is the script that runs the tests and generates the reports:

Here is the console output.

Here is the generated report.

Please, let me know if you need further information.

[danielpalme]

ReportGenerator counts branches in the following way:

For each line: Get number of branches from XML For each file file: Sum of all lines For each class: Sum of all files For each assembly: Sum of all classes Total: Sum of all assemblies

Only "real" branches ( in OpenCover XML) are taken into account. OpenCover counts differently here, it counts one extra branch per method.

See also: https://github.com/danielpalme/ReportGenerator/issues/236 https://github.com/danielpalme/ReportGenerator/issues/191

[stoyandimov]

Thanks for the explanation.