Needle editor is very unreliable when the worker is run without cleanups
I've had to debug a test today that has a relatively expensive setup in an extra step and failed somewhere in the middle. For that, I wanted to utilize test snapshots as described here: https://github.com/os-autoinst/openQA/blob/master/docs/WritingTests.asciidoc#using-snapshots-to-speed-up-development-of-tests
While they work perfectly fine, running the worker without cleanup makes the needle editor completely unusable (which I had to use after finding the issue). It selects arbitrary screenshots that it made at some point as the last one. That makes creating new needles pretty much impossible.
It selects arbitrary screenshots that it made at some point as the last one.
That's strange. Maybe the worker attempts to upload screenshots from the previous run here which get mixed into the pod of screenshots the needle editor works with and it can not handle it well. That would mean the worker is at fault here because it should not upload screenshots from a previous run. If that theory is true, you might be able to workaround the issue by deleting all PNGs within the pool directory manually before starting the 2nd run.
There's another workaround: You can start the needle editor from any screenshot shown within the details tab. This way is should be possible to select the screenshot you would like to base your new needle on manually. Or are you doing this already and the needle editor is so messed up that it doesn't even show the right screenshot if you open it from a specific step/screenshot?