@pvdrz I'm a bit on the fence on whether #2787 should be considered a breaking change or not. On one hand, it can make code that previously compiled (but failed unit tests) fail to build. On the other, those are broken bindings. Do you have a strong feeling either way?
@emilio I'd say it is a breaking change anyway. Even if the bindings wouldn't compile on the host, someone could be taking snapshots of their bindings to avoid breaking changes themselves.
@pvdrz I'm a bit on the fence on whether #2787 should be considered a breaking change or not. On one hand, it can make code that previously compiled (but failed unit tests) fail to build. On the other, those are broken bindings. Do you have a strong feeling either way?