Closed aloucks closed 4 years ago
@kvark I don't think bors is hooked up (based on your previous comments on another PR)
I wonder if this can be considered a breaking change...
I think this is a breaking change iff sufficiently strong guarantees are asserted by the documentation to permit nontrivial sound transmute
s. There's a reasonable argument that #[repr(C)]
, which shows up in rustdoc, constitutes such a guarantee. On the other hand, I suspect nobody will be affected, and it'd be nice to not break everything.
I think we can take that risk, with the gain of disrupting less of existing "safe" users. Ready to yank it if reports prove that to be a bad idea :) r? https://github.com/kvark/mint/pull/53 @aloucks @Ralith
Is there any downside to bumping the minor version?
EDIT: Ah, I saw this before I saw it was released. Nevermind about the version!
The Quaternion memory layout is now
[x, y, z, w]
(i.e. scalar at the end). This is consistent with theInto
trait implementation and also consistent with other libraries (such as GLM) and the GLTF spec.Fixes #49