Open Bug 1821522 Opened 2 years ago Updated 1 year ago

investigate how to use single tracking bugs for leaks/crashes

Categories

(Testing :: General, task)

Default
task

Tracking

(Not tracked)

People

(Reporter: jmaher, Unassigned)

References

(Blocks 5 open bugs)

Details

typically we get leaks and crashes and process them after the browser has terminated. Sometimes we have a test case, other times we know the manifest that was run previously.

Ideally we would end up with a way to track the leak | crash, and then use the test|manifest as the error, so currently a failing test has this format:

test-unexpected-fail | <testcase> | <error>

and I propose something like:
test-unexpected-fail | leak string | <manifest>
test-unexpected-fail | crash string | <manifest>

this would allow us to file a bug that looks like:
[meta] intermittent | leak string | single tracking bug

then tools like intermittent failure view and mach could help summarize the <manifest> that caused the problem.

this would take some research into what a useful leak/crash string would be. Also what edge cases might exist and if there are different reporting of crash/leak in browser/harness.

Blocks: 1821525
Blocks: 1821527
Blocks: 1821528
Blocks: 1821531

currently we are doing this for crashes, leaks and assertions need some attention, but it should be easy to follow along. I will leave this open until those are done. Assuming regular working in August (as planned) I am looking to put 2-3 days into single tracking bugs with the goal if getting this ready for the final stage (meta bugs, and resolving this and other bugs)

You need to log in before you can comment on or make changes to this bug.