Open gmazzo opened 1 year ago
I am currently facing the same issue with test-application
. Whenever there is a failing test the steps errors and all the outputs are null.
I can see here that you throw an error without setting any outputs.
if (run_status.run.result != "PASSED") {
throw("Test run failed after " + i*wait_interval + " seconds with: " + run_status.run.resultCode + ". Timeout is set to " + max_wait*wait_interval);
}
I can see here that you throw an error without setting any outputs.
I'll revisit this soon - at the time of writing this I didn't get any meaningful reply to be returned here from the devicefarm API, and therefore just had it fail without any outputs.
I can see here that you throw an error without setting any outputs.
I'll revisit this soon - at the time of writing this I didn't get any meaningful reply to be returned here from the devicefarm API, and therefore just had it fail without any outputs.
@bwachter Thanks for your reply.
In our case the run arn would be very useful. It is already available straight after the schedule run is called. In this case when using the test-application
and the run fails we can show a link to the run
In our case the run arn would be very useful. It is already available straight after the schedule run is called. In this case when using the
test-application
and the run fails we can show a link to the run
That's a good point - and at minimum I'll add that when I work on this again.
Assumption when building this was that developers calling this typically don't have access to the device farm - so the focus was on pulling enough information to debug errors from the API and output them. Now with more 3rd parties using this action I might have to revisit some of those initial assumptions.
I have added the proposed change here
I'm currently using
${{ steps.adf.outputs.arn }}
withto reconstruct the test run URL to provide feedback to the users. But it seems when the step fails, all
outputs
are empty