Adrian Taylor

@johnperry-math I'd like to land this test now with an extra attribute: `#[ignore] // https://github.com/google/autocxx/issues/774` to avoid future merge conflicts, like the one which just occurred. Please could you update...

Thanks, but I fear you added the attribute to the wrong test!

A (presumably) flakey test: https://github.com/google/autocxx/runs/5337017794?check_suite_focus=true - `test_uniqueptr_moveit`. This might be a real bug which just happens to manifest on Windows only due to ABI issues. I'd like to figure out...

Ah, it was kind of amazing that nobody spotted bugs in that constructor-related logic until now. Good spot.

@youben11 I don't think this is believed to be fixed yet.

I think all you can do is explicitly `block!(mlir::DiagnosticEngine::DiagnosticEngine)` but I appreciate this may make autocxx useless for your needs.

Some work in progress: * #1180 * https://github.com/adetaylor/rust-bindgen/compare/master...adetaylor:rust-bindgen:denote-defaulted These two PRs don't yet connect together: although autocxx now knows about defaulted functions, it doesn't treat them any differently than it...

Thanks, this is definitely a bug in autocxx. Would you mind raising a PR with a [failing test case](https://google.github.io/autocxx/contributing.html#reporting-bugs)?

Sorry for the delay on this. This is next on my list of bugs to take a look at.

The problem is that the contents of the vector is a forward-declared type, I think: ``` ../../buildtools/third_party/libc++/trunk/include/vector:659:53: error: arithmetic on a pointer to an incomplete type 'content::AXLocationChangeNotificationDetails' {return static_cast(this->__end_ -...