We only check the contents of the debug output files for the regression test for the simple example (to avoid unnecessary churn associated with regenerating them all the time), but they are useful for debugging (hence the name!) so it is probably a good idea to generate these files for every test, even if they're not checked. As the output files are uploaded to GitHub as test artifacts, having this extra info could be useful for debugging failing tests (see e.g. #1185).
We should probably ensure that we are still left with at least one integration test with the debug_model setting disabled though (maybe we should add another for simple?).
We only check the contents of the debug output files for the regression test for the
simpleexample (to avoid unnecessary churn associated with regenerating them all the time), but they are useful for debugging (hence the name!) so it is probably a good idea to generate these files for every test, even if they're not checked. As the output files are uploaded to GitHub as test artifacts, having this extra info could be useful for debugging failing tests (see e.g. #1185).We should probably ensure that we are still left with at least one integration test with the
debug_modelsetting disabled though (maybe we should add another forsimple?).