In https://bugzilla.mozilla.org/show_bug.cgi?id=1632870, we added a new test-setting property to the extra section of test tasks. We should parse this to obtain platform / build / runtime properties of test tasks rather than relying on the label. E.g:
|
def get_configuration_from_label(label: str) -> str: |
We should also provide an abstraction for settings, and a way to group tasks by any arbitrary subset of the setting. This will unlock some exciting possibilities, like finer grained selection of tasks in the bugbug ML, better dashboards in Orangefactor, a way for a tool to try to identify commonalities of intermittents, etc.