The run for 2089041 finished and the differences aren't massive:
https://wpt.fyi/results/?q=is%3Adifferent&run_id=627220001&run_id=648510001

I have pushed 78424f2 to triggers/safari_preview. The nearest parent is 6ceb598 which is unfortunately a few commits back but should be roughly ok for a comparison.

Once the run has finished I will post the wpt.fyi comparison url.

cc @ziransun

https://wpt.fyi/results/?diff&filter=ADC&run_id=694810001&run_id=702810004 is the diff, but there's something strange there. Multiple directories have hundreds of tests now MISSING, and unless someone deleted them in 6ceb598...78424f2 (which I doubt) then something is wrong with the run.

@ziransun FYI as you might want to take a look anyway, but really I need to dig into the Azure Pipelines logs for the run.

I wasn't able to find anything obvious in the logs at a quick look, so I'm being lazy and just triggering another run to see. eb419a7 is the run, the parent is c960a24 aka https://wpt.fyi/results/?run_id=683470005

Ok, the new run seems much more reasonable: https://wpt.fyi/results/?diff&filter=ADC&run_id=683470005&run_id=695000002 . A little concerning to not figure out what was happening last time, but for now I think we can proceed with using this as the newest Safari run.

cc @ziransun