Pytest's default JUnit format in 6.0 changed from "legacy" to "xunit2" (source). 6.0 shipped in 2020-07-28.
In the current "xunit2" format, the generated XML report is missing the "file" attribute. This is a problem for us, because we rely on this file information in order to construct test path. A crucial bit of information to be able to do subsetting. It means we are not compatible with pytest out of the box.
This is documented, which is great, but I don't think that's sufficient.
Currently, CLI happily eats this new report file without any errors. The resulting test path doesn't include the file attribute, which breaks subsetting.
This section talks about the accepted "node ID" format, which requires the file name.
Pytest's default JUnit format in 6.0 changed from "legacy" to "xunit2" (source). 6.0 shipped in 2020-07-28.
In the current "xunit2" format, the generated XML report is missing the "file" attribute. This is a problem for us, because we rely on this file information in order to construct test path. A crucial bit of information to be able to do subsetting. It means we are not compatible with pytest out of the box.
This is documented, which is great, but I don't think that's sufficient.
Currently, CLI happily eats this new report file without any errors. The resulting test path doesn't include the file attribute, which breaks subsetting.
This section talks about the accepted "node ID" format, which requires the file name.