Closed kelloggm closed 3 months ago
If this passes CI, I will override approval requirements and merge it without review from @LoiNguyenCS. He should review it later when he gets a chance.
I just assumed that removing the unused type bound should not affect an analysis tool.
No worries, I should have noticed earlier. This is why we weren't preserving one of the JDK bugs, though, so I wanted to fix it (there was no easy way to sell "our modularity model is wrong" in the paper).
@LoiNguyenCS next week you should look through the other changes I made in this PR to keep CF-6388 compiling. They're pretty extensive; we had the modularity model for interfaces wrong, too.
This should be part of the modularity model. The previous test case that showed we didn't do this was wrong, and I should not have approved it.
Should fix JDK-8319461.