Parry benchmarks compilation #255
Conversation
| @@ -143,7 +142,7 @@ bench_method!( | |||
| bounding_sphere: BoundingSphere, | |||
| c: ConvexHull, | |||
| m: Isometry3<f32> | |||
| ); | |||
| );*/ | |||
There was a problem hiding this comment.
ConvexHull is no longer a type as iitself, but rather an alias to Vec<Point<Real>>, I think it would have value to be its own type, but slightly out of scope for this PR.
@sebcrozet I'd appreciate instructions as how to proceed:
Should I straight up remove that code, or keep it in and create a tracking issue (+ comment) ?
Then, should I either:
- make it work with the alias ?
- implement the methods for
Vec<Point<Real>>, which will probably be confusing
- implement the methods for
- make a
ConvexHulltype. An opportunity for better docs in my opinion,ConvexHullexists in the current documentation at a few places ([here](i.e: https://github.com/Vrixyz/parry/blob/7e76ebe0c7a155e44d631d4f25a227cf7daf598c/src/query/closest_points/closest_points_support_map_support_map.rs#L8) or there...)
There was a problem hiding this comment.
The ConvexHull shpe type has been renamed ConvexPolyhedron in 3D and ConvexPolygon in 2D.
crates/parry3d/benches/all.rs
Outdated
|
|
||
| #[cfg(feature = "dim2")] | ||
| type ConvexHull = Vec<Point<Real>>; | ||
| #[cfg(feature = "dim3")] | ||
| type ConvexHull = (Vec<Point<Real>>, Vec<[u32; 3]>); |
There was a problem hiding this comment.
| #[cfg(feature = "dim2")] | |
| type ConvexHull = Vec<Point<Real>>; | |
| #[cfg(feature = "dim3")] | |
| type ConvexHull = (Vec<Point<Real>>, Vec<[u32; 3]>); |
| @@ -143,7 +142,7 @@ bench_method!( | |||
| bounding_sphere: BoundingSphere, | |||
| c: ConvexHull, | |||
| m: Isometry3<f32> | |||
| ); | |||
| );*/ | |||
There was a problem hiding this comment.
The ConvexHull shpe type has been renamed ConvexPolyhedron in 3D and ConvexPolygon in 2D.
| @@ -72,7 +72,7 @@ macro_rules! bench_method_gen ( | |||
|
|
|||
| unsafe { | |||
| let val: $tres = test::black_box($arg.get_unchecked(i).$method($(unref($args.get_unchecked(i)),)*)); | |||
| drop(val); | |||
| let _ = val; | |||
There was a problem hiding this comment.
using drop results in a warning:
warning: calls to `std::mem::drop` with a value that implements `Copy` does nothing
--> crates/parry3d/benches/bounding_volume/../common/macros.rs:75:21Aabb and BoundingSphere are concerned. I don´t think this is an opportunity to make benchmarks more accurate (if we're unexpectedly copying).
Actually I'm not exactly sure why this drop is necessary. As we're in an unsafe call, I guess it might be useful, but not really used in current usage. It doesn´t hurt being here I guess.
Fixes compiling for parry benchmarks
There has been a few API changes since their creation, especially around
ConvexHull, and in a lesser amountCapsule.