You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently the logging behaviour for "debug" level logs is set to show them at the end of the integration test, if there were any failures. This results in them not showing up at all, should the integration test prematurely exit to due a crash, failed assertion, or timeout. In practice, this often leads to a lack of relevant logs to understand why a test didn't complete.
To fix this, the overall log level is changed from info to debug. This ensures debug logs are logged immediately, so a prematurely ended test won't prevent them from showing up.
This also includes a small fix to the desktop_tester script to capture logs in the event of a process timeout: previously, result.stdout was used, which doesn't get populated if the process timed out. Instead, we access output from the error. Experimentation revealed that this output will sometimes be test, and sometimes binary: it's unclear what's responsible for this difference (possibly the encoding settings of the environment running the script). Thus we try to decode, and store it as-is if that fails.
0 commit comments