Re: noverlay branch

[Stefan Monnier]

>>      node->otick = otick;
>
> Indeed, I saw later in the remove code that we do propagate offsets
> without caring if they're up-to-date, and I think the logic behind that
> is sound.  And indeed we can drop the test because the only property we
> only ever care about for an `otick` is whether it's equal to
> `tree->otick`, so if node->parent->otick != otick then performing the
> assignment is equivalent to not performing it.

The above isn't quite right:
I was thinking of `node->otick = parent->otick`.


        Stefan