Local execution of long tests show that , following a refactoring, some tests now fail. These tests make use of some trends that are not enabled by default (moving average, moving median).
Add a set of tests to run at least each model component once in the minimal set of build tests.
Now at least 280 test scripts producing about 1000 models are built on each circleci run. There is still a limit of 1 hour on the whole run.
Local execution of long tests show that , following a refactoring, some tests now fail. These tests make use of some trends that are not enabled by default (moving average, moving median).
Add a set of tests to run at least each model component once in the minimal set of build tests.
Now at least 280 test scripts producing about 1000 models are built on each circleci run. There is still a limit of 1 hour on the whole run.