Thanks. Yeah, that's pretty much what I thought was going on. I've spent a bunch of time combing through my code and the memory graph, converting lots of reference to weak, but the leak remains. One thing that confuses me is when I see an object in the memory graph (such as a view controller) where all of the references to it are "conservative", which seems to mean "weak". If they're all weak, why does the object remain?

Ah, that was the issue - I grabbed the session off the App instead of a view. Moving it down to a view solves the issue. Thanks!

Yes, it's iOS and we always run our tests in a simulator.

@DTS Engineer I tried that, and it doesn't seem to have any effect on the accessibility elements I see. The identifier still ends up getting applied to the static text instead.

@DTS Engineer A good example test is here: https://github.com/Uncommon/Xit/blob/wip-cellID/XitUITests/ModifyingUITests.swift#L294 The view is here: https://github.com/Uncommon/Xit/blob/wip-cellID/Xit/Sidebar/BranchList.swift#L65

@DTS Engineer Thanks for the simplified example. It's interesting that you're referencing app.collectionViews. For my case, it shows up under outlines even if I only have the first cell. Does that make a difference for what I'm trying to do? Also, your test verifies that the "StagingCell" ID is getting set on the static text, which is the result I am currently getting. Can I get the ID set on the cell itself instead?

Did you see the link in my reply below? I thought I had read that you get notified for replies but not comments. Maybe I had it backward.