# Objective
- Avoid unnecessary work in the vertex shader of the numerous shadow passes
- Have the natural order of bind groups in the pbr shader: view, material, mesh
## Solution
- Separate out the vertex stage of pbr.wgsl into depth.wgsl
- Remove the unnecessary calculation of uv and normal, as well as removing the unnecessary vertex inputs and outputs
- Use the depth.wgsl for shadow passes
- Reorder the bind groups in pbr.wgsl and PbrShaders to be 0 - view, 1 - material, 2 - mesh in decreasing order of rebind frequency
# Objective
Allow marking meshes as not casting / receiving shadows.
## Solution
- Added `NotShadowCaster` and `NotShadowReceiver` zero-sized type components.
- Extract these components into `bool`s in `ExtractedMesh`
- Only generate `DrawShadowMesh` `Drawable`s for meshes _without_ `NotShadowCaster`
- Add a `u32` bit `flags` member to `MeshUniform` with one flag indicating whether the mesh is a shadow receiver
- If a mesh does _not_ have the `NotShadowReceiver` component, then it is a shadow receiver, and so the bit in the `MeshUniform` is set, otherwise it is not set.
- Added an example illustrating the functionality.
NOTE: I wanted to have the default state of a mesh as being a shadow caster and shadow receiver, hence the `Not*` components. However, I am on the fence about this. I don't want to have a negative performance impact, nor have people wondering why their custom meshes don't have shadows because they forgot to add `ShadowCaster` and `ShadowReceiver` components, but I also really don't like the double negatives the `Not*` approach incurs. What do you think?
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
Add support for configurable shadow map sizes
## Solution
- Add `DirectionalLightShadowMap` and `PointLightShadowMap` resources, which just have size members, to the app world, and add `Extracted*` counterparts to the render world
- Use the configured sizes when rendering shadow maps
- Default sizes remain the same - 4096 for directional light shadow maps, 1024 for point light shadow maps (which are cube maps so 6 faces at 1024x1024 per light)
Makes some tweaks to the SubApp labeling introduced in #2695:
* Ergonomics improvements
* Removes unnecessary allocation when retrieving subapp label
* Removes the newly added "app macros" crate in favor of bevy_derive
* renamed RenderSubApp to RenderApp
@zicklag (for reference)
# Objective
- Clarify vague meaning of "Ltr" and "Rtl". For someone familiar with Flex Box, this is easy to understand, but being more explicit will help beginners or those unfamiliar, without the need to do research.
## Solution
- Change three letter abbreviation to fully descriptive name.
This matches `ahash::RandomState`, which provides both `Debug` and `Clone`.
Notably, implementing `Clone` allows the `StableHashMap`/`Set` to also implement `Clone`.
# Objective
- Allow `bevy_utils::StableHashMap` to be cloned.
## Solution
- Derive `Clone` for `bevy_utils::FixedState`.
- Also derive `Debug`, since we're touching it anyway, and this aligns `FixedState` with `ahash::RandomState`.
This is a rather simple but wide change, and it involves adding a new `bevy_app_macros` crate. Let me know if there is a better way to do any of this!
---
# Objective
- Allow adding and accessing sub-apps by using a label instead of an index
## Solution
- Migrate the bevy label implementation and derive code to the `bevy_utils` and `bevy_macro_utils` crates and then add a new `SubAppLabel` trait to the `bevy_app` crate that is used when adding or getting a sub-app from an app.
# Objective
A question was raised on Discord about the units of the `PointLight` `intensity` member.
After digging around in the bevy_pbr2 source code and [Google Filament documentation](https://google.github.io/filament/Filament.html#mjx-eqn-pointLightLuminousPower) I discovered that the intention by Filament was that the 'intensity' value for point lights would be in lumens. This makes a lot of sense as these are quite relatable units given basically all light bulbs I've seen sold over the past years are rated in lumens as people move away from thinking about how bright a bulb is relative to a non-halogen incandescent bulb.
However, it seems that the derivation of the conversion between luminous power (lumens, denoted `Φ` in the Filament formulae) and luminous intensity (lumens per steradian, `I` in the Filament formulae) was missed and I can see why as it is tucked right under equation 58 at the link above. As such, while the formula states that for a point light, `I = Φ / 4 π` we have been using `intensity` as if it were luminous intensity `I`.
Before this PR, the intensity field is luminous intensity in lumens per steradian. After this PR, the intensity field is luminous power in lumens, [as suggested by Filament](https://google.github.io/filament/Filament.html#table_lighttypesunits) (unfortunately the link jumps to the table's caption so scroll up to see the actual table).
I appreciate that it may be confusing to call this an intensity, but I think this is intended as more of a non-scientific, human-relatable general term with a bit of hand waving so that most light types can just have an intensity field and for most of them it works in the same way or at least with some relatable value. I'm inclined to think this is reasonable rather than throwing terms like luminous power, luminous intensity, blah at users.
## Solution
- Documented the `PointLight` `intensity` member as 'luminous power' in units of lumens.
- Added a table of examples relating from various types of household lighting to lumen values.
- Added in the mapping from luminous power to luminous intensity when premultiplying the intensity into the colour before it is made into a graphics uniform.
- Updated the documentation in `pbr.wgsl` to clarify the earlier confusion about the missing `/ 4 π`.
- Bumped the intensity of the point lights in `3d_scene_pipelined` to 1600 lumens.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
The default perspective projection near plane being at 1 unit feels very far away if one considers units to directly map to real world units such as metres. Not being able to see anything that is closer than 1m is unnecessarily limiting. Using a default of 0.1 makes more sense as it is difficult to even focus on things closer than 10cm in the real world.
## Solution
- Changed the default perspective projection near plane to 0.1.
# Objective
- Allow the user to set the clear color when using the pipelined renderer
## Solution
- Add a `ClearColor` resource that can be added to the world to configure the clear color
## Remaining Issues
Currently the `ClearColor` resource is cloned from the app world to the render world every frame. There are two ways I can think of around this:
1. Figure out why `app_world.is_resource_changed::<ClearColor>()` always returns `true` in the `extract` step and fix it so that we are only updating the resource when it changes
2. Require the users to add the `ClearColor` resource to the render sub-app instead of the parent app. This is currently sub-optimal until we have labled sub-apps, and probably a helper funciton on `App` such as `app.with_sub_app(RenderApp, |app| { ... })`. Even if we had that, I think it would be more than we want the user to have to think about. They shouldn't have to know about the render sub-app I don't think.
I think the first option is the best, but I could really use some help figuring out the nuance of why `is_resource_changed` is always returning true in that context.
# Objective
- the plugin guidelines should be up-to-date and easy to read/understand
## Solution
* point to "Bevy Assets" instead of old "Awesome Bevy"
* restructure sections
* same order for sections and checklist
* Update examples with newest release/rev
Updates the requirements on [glam](https://github.com/bitshifter/glam-rs) to permit the latest version.
<details>
<summary>Changelog</summary>
<p><em>Sourced from <a href="https://github.com/bitshifter/glam-rs/blob/master/CHANGELOG.md">glam's changelog</a>.</em></p>
<blockquote>
<h2>[0.17.3] - 2021-07-18</h2>
<h3>Fixed</h3>
<ul>
<li>Fix alignment unit tests on non x86 platforms.</li>
</ul>
<h2>[0.17.2] - 2021-07-15</h2>
<h3>Fixed</h3>
<ul>
<li>Fix alignment unit tests on i686 and S390x.</li>
</ul>
<h2>[0.17.1] - 2021-06-29</h2>
<h3>Added</h3>
<ul>
<li>Added <code>serde</code> support for <code>Affine2</code>, <code>DAffine2</code>, <code>Affine3A</code> and <code>DAffine3</code>.</li>
</ul>
<h2>[0.17.0] - 2021-06-26</h2>
<h3>Breaking changes</h3>
<ul>
<li>The addition of <code>Add</code> and <code>Sub</code> implementations of scalar values for vector
types may create ambiguities with existing calls to <code>add</code> and <code>sub</code>.</li>
<li>Removed <code>From<Mat3></code> implementation for <code>Mat2</code> and <code>From<DMat3></code> for <code>DMat2</code>.
These have been replaced by <code>Mat2::from_mat3()</code> and <code>DMat2::from_mat3()</code>.</li>
<li>Removed <code>From<Mat4></code> implementation for <code>Mat3</code> and <code>From<DMat4></code> for <code>DMat3</code>.
These have been replaced by <code>Mat3::from_mat4()</code> and <code>DMat3::from_mat4()</code>.</li>
<li>Removed deprecated <code>from_slice_unaligned()</code>, <code>write_to_slice_unaligned()</code>,
<code>from_rotation_mat4</code> and <code>from_rotation_ypr()</code> methods.</li>
</ul>
<h3>Added</h3>
<ul>
<li>Added <code>col_mut()</code> method which returns a mutable reference to a matrix column
to all matrix types.</li>
<li>Added <code>AddAssign</code>, <code>MulAssign</code> and <code>SubAssign</code> implementations for all matrix
types.</li>
<li>Added <code>Add</code> and <code>Sub</code> implementations of scalar values for vector types.</li>
<li>Added more <code>glam_assert!</code> checks and documented methods where they are used.</li>
<li>Added vector projection and rejection methods <code>project_onto()</code>,
<code>project_onto_normalized()</code>, <code>reject_from()</code> and <code>reject_from_normalized()</code>.</li>
<li>Added <code>Mat2::from_mat3()</code>, <code>DMat2::from_mat3()</code>, <code>Mat3::from_mat4()</code>,
<code>DMat3::from_mat4()</code> which create a smaller matrix from a larger one,
discarding a final row and column of the input matrix.</li>
<li>Added <code>Mat3::from_mat2()</code>, <code>DMat3::from_mat2()</code>, <code>Mat4::from_mat3()</code> and
<code>DMat4::from_mat3()</code> which create an affine transform from a smaller linear
transform matrix.</li>
</ul>
<h3>Changed</h3>
<!-- raw HTML omitted -->
</blockquote>
<p>... (truncated)</p>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a href="ecf3904b2f"><code>ecf3904</code></a> Prepare release 0.17.3</li>
<li><a href="95e02bb43e"><code>95e02bb</code></a> Merge branch 'master' of github.com:bitshifter/glam-rs</li>
<li><a href="c6dc702583"><code>c6dc702</code></a> More alignment test fixes for when SSE2 is not avaialable.</li>
<li><a href="87a3b25872"><code>87a3b25</code></a> Merge pull request <a href="https://github-redirect.dependabot.com/bitshifter/glam-rs/issues/216">#216</a> from bitshifter/prepare-0.17.2</li>
<li><a href="269e514090"><code>269e514</code></a> Prepare for 0.17.2 release.</li>
<li><a href="1da7d6459c"><code>1da7d64</code></a> Merge pull request <a href="https://github-redirect.dependabot.com/bitshifter/glam-rs/issues/215">#215</a> from bitshifter/issue-213</li>
<li><a href="dc60e20925"><code>dc60e20</code></a> Fix align asserts on i686 and S390x architectures.</li>
<li><a href="bd8b30e9fb"><code>bd8b30e</code></a> Merge pull request <a href="https://github-redirect.dependabot.com/bitshifter/glam-rs/issues/212">#212</a> from remilauzier/master</li>
<li><a href="a4e97c0b54"><code>a4e97c0</code></a> Update approx to 0.5</li>
<li><a href="059f619525"><code>059f619</code></a> Prepare 0.17.1 release (<a href="https://github-redirect.dependabot.com/bitshifter/glam-rs/issues/211">#211</a>)</li>
<li>Additional commits viewable in <a href="https://github.com/bitshifter/glam-rs/compare/0.15.1...0.17.3">compare view</a></li>
</ul>
</details>
<br />
Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.
[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)
---
<details>
<summary>Dependabot commands and options</summary>
<br />
You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually
- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
</details>
# Objective
- We currently depends on ndk 0.2, 0.3, 0.4
- Only 0.2 dependencies comes from Bevy itself
## Solution
- Replace #1371
- Update Bevy to ndk-glue 0.4
- Also fixes duplicate dependency CI issue
# Objective
While implementing a plugin for my rollback networking library, I needed to load/save parts of the world. For this, I made a WorldSnapshot that works quite like the current DynamicScene. Using a TypeRegistry to register component types I want to save/load and then using ReflectComponents methods to add or apply components of the given types.
However, I noticed there is no method to remove components from entities through the ReflectComponent.
## Solution
I added a `remove_component` field to the `ReflectComponent` struct, as well as a `pub fn remove_component(&self, world: &mut World, entity: Entity)` to call that function in `remove_component`. This follows exactly the same pattern all other methods/fields in this struct look like.
This is an example how it could be used (at least how I would use it):
6c003f86f1/src/world_snapshot.rs (L133)
# Objective
Enable using exact World lifetimes during read-only access . This is motivated by the new renderer's need to allow read-only world-only queries to outlive the query itself (but still be constrained by the world lifetime).
For example:
115b170d1f/pipelined/bevy_pbr2/src/render/mod.rs (L774)
## Solution
Split out SystemParam state and world lifetimes and pipe those lifetimes up to read-only Query ops (and add into_inner for Res). According to every safety test I've run so far (except one), this is safe (see the temporary safety test commit). Note that changing the mutable variants to the new lifetimes would allow aliased mutable pointers (try doing that to see how it affects the temporary safety tests).
The new state lifetime on SystemParam does make `#[derive(SystemParam)]` more cumbersome (the current impl requires PhantomData if you don't use both lifetimes). We can make this better by detecting whether or not a lifetime is used in the derive and adjusting accordingly, but that should probably be done in its own pr.
## Why is this a draft?
The new lifetimes break QuerySet safety in one very specific case (see the query_set system in system_safety_test). We need to solve this before we can use the lifetimes given.
This is due to the fact that QuerySet is just a wrapper over Query, which now relies on world lifetimes instead of `&self` lifetimes to prevent aliasing (but in systems, each Query has its own implied lifetime, not a centralized world lifetime). I believe the fix is to rewrite QuerySet to have its own World lifetime (and own the internal reference). This will complicate the impl a bit, but I think it is doable. I'm curious if anyone else has better ideas.
Personally, I think these new lifetimes need to happen. We've gotta have a way to directly tie read-only World queries to the World lifetime. The new renderer is the first place this has come up, but I doubt it will be the last. Worst case scenario we can come up with a second `WorldLifetimeQuery<Q, F = ()>` parameter to enable these read-only scenarios, but I'd rather not add another type to the type zoo.
[**RENDERED**](https://github.com/alice-i-cecile/bevy/blob/better-contributing/CONTRIBUTING.md)
Improves #910. As discussed in #1309, we'll need to synchronize content between this and the Bevy website in some way (and clean up the .github file perhaps?).
I think doing it as a root-directory file is nicer for discovery, but that's a conversation I'm interested in having.
This document is intended to be helpful to beginners to open source and Bevy, and captures what I've learned about our informal practices and values.
Reviewers: I'm particularly interested in:
- opinions on the items **What we're trying to build**, where I discuss some of the project's high-level values and goals
- more relevant details on the `bevy` subcrates for **Getting oriented**
- useful tricks and best practices that I missed
- better guidance on how to contribute to the Bevy book from @cart <3
# Objective
This:
```rust
use bevy::prelude::*;
fn main() {
App::new()
.add_system(test)
.run();
}
fn test(entities: Query<Entity>) {
let mut combinations = entities.iter_combinations_mut();
while let Some([e1, e2]) = combinations.fetch_next() {
dbg!(e1);
}
}
```
fails with the message "the trait bound `bevy::ecs::query::EntityFetch: std::clone::Clone` is not satisfied".
## Solution
It works after adding the naive clone implementation to EntityFetch. I'm not super familiar with ECS internals, so I'd appreciate input on this.
## Objective
- Clean up remaining references to the trait `FromResources`, which was replaced in favor of `FromWorld` during the ECS rework.
## Solution
- Remove the derive macro for `FromResources`
- Change doc references of `FromResources` to `FromWorld`
(this is the first item in #2576)
# Objective
- Provides more useful error messages when using unsupported shader features.
## Solution Fixes#869
- Provided a error message as follows (adding name, set and binding):
```
Unsupported shader bind type CombinedImageSampler (name noiseVol0, set 0, binding 9)
```
# Objective
Fix ComputePipelineDescriptor missing from WGPU exports
## Solution
Added it to the pub use wgpu::{ ... }
Co-authored-by: Dimas <skythedragon@outlook.com>
This is an updated version of #1434 PR. I've encountered this macro problem while trying to use @woubuc's bevy-event-set crate.
Co-authored-by: Piotr Balcer <piotr@balcer.eu>
I didn't know about MinimalPlugins for way too long. This should increase visibility for others.
# Objective
Improve visibility and discover in the docs for Default and Minimal Plugins.
## Solution
Links the two Docs pages.
Co-authored-by: Mirko Rainer <52899592+mirkoRainer@users.noreply.github.com>
# Objective
- Allow `ScheduleRunnerPlugin` to be instantiated without curly braces. Other plugins in the library already use the semicolon syntax.
- Currently, you have to do the following:
```rust
App::build()
.add_plugin(bevy::core::CorePlugin)
.add_plugin(bevy::app::ScheduleRunnerPlugin {})
```
- With the proposed change you can do this:
```rust
App::build()
.add_plugin(bevy::core::CorePlugin)
.add_plugin(bevy::app::ScheduleRunnerPlugin)
```
## Solution
- Change the `ScheduleRunnerPlugin` definition to use a semicolon instead of curly braces.
# Objective
- Prevent the need to specify a sprite size when using the pipelined sprite renderer
## Solution
- Re-introduce the sprite auto resize system from the old renderer
# Objective
- Allow you to compile Bevy with the `bevy_sprite2` feature, but without the `bevy_pbr2` feature.
- This currently fails because the `bevy_sprite2` crate does not require the `derive` feature of the `bytemuck` crate in its `Cargo.toml`, even though it is required to compile.
## Solution
- Add the `derive` feature of `bytemuck` to the `bevy_sprite2` crate
## Objective
This code would result in a crash:
```rust
use bevy::prelude::*;
fn main() {
let mut world = World::new();
let child = world.spawn().id();
world.spawn().push_children(&[child]);
}
```
## Solution
Update the `EntityMut`'s location after inserting a component on the children entities, as it may have changed.
# Objective
This fixes not having access to StorageTextureAccess in the API, which is needed for using storage textures
## Solution
Added it to the use in render_resource module
Co-authored-by: Dimas <skythedragon@outlook.com>
# Objective
- #2551 revamped our CI setup which included running clippy and rustfmt in another Job.
- This new Job wasn't added to the bors.toml, which means that PRs would be accepted that didn't run them.
## Solution
- Add the "ci" job to the bors.toml
# Objective
Restore the functionality of sprite atlases in the new renderer.
### **Note:** This PR relies on #2555
## Solution
Mostly just a copy paste of the existing sprite atlas implementation, however I unified the rendering between sprites and atlases.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
It doesn't compile on wasm, and it's full of footguns
# Objective
- If bevy is used with default features on wasm, there's more of a chance it will compile
- Note that I haven't done a full audit - it's possible that there are other problematic crates
## Solution
- `bevy_dynamic_plugin` is no longer a default plugin
- I've also done an accidental drive by reformatting of the root `Cargo.toml`, as I have [Even Better Toml](https://github.com/tamasfe/taplo) installed.
- (Please, rustfmt do this for us)
# Objective
Port bevy_gltf to the pipelined-rendering branch.
## Solution
crates/bevy_gltf has been copied and pasted into pipelined/bevy_gltf2 and modifications were made to work with the pipelined-rendering branch. Notably vertex tangents and vertex colours are not supported.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
notify 5.0.0-pre.11 breaks the interface again, but apparently in a way that's similar to how it used to be
## Solution
Bump `bevy_asset` dependency on notify to `5.0.0-pre.11` and fix the errors that crop up.
It looks like `pre.11` was mentioned in #2528 by @mockersf but there's no mention of why `pre.10` was chosen ultimately.
This decouples the opinionated "core pipeline" from the new (less opinionated) bevy_render crate. The "core pipeline" is intended to be used by crates like bevy_sprites, bevy_pbr, bevy_ui, and 3rd party crates that extends core rendering functionality.
# Objective
There is currently a 1-to-1 mapping between components and real rust types. This means that it is impossible for multiple components to be represented by the same rust type or for a component to not have a rust type at all. This means that component types can't be defined in languages other than rust like necessary for scripting or sandboxed (wasm?) plugins.
## Solution
Refactor `ComponentDescriptor` and `Bundle` to remove `TypeInfo`. `Bundle` now uses `ComponentId` instead. `ComponentDescriptor` is now always created from a rust type instead of through the `TypeInfo` indirection. A future PR may make it possible to construct a `ComponentDescriptor` from it's fields without a rust type being involved.
Makes the "Render App World" directly available to Extract step systems as a `RenderWorld` resource. Prior to this, there was no way to directly read / write render world state during the Extract step. The only way to make changes was through Commands (which were applied at the end of the stage).
```rust
// `thing` is an "app world resource".
fn extract_thing(thing: Res<Thing>, mut render_world: ResMut<RenderWorld>) {
render_world.insert_resource(ExtractedThing::from(thing));
}
```
RenderWorld makes a number of scenarios possible:
* When an extract system does big allocations, it is now possible to reuse them across frames by retrieving old values from RenderWorld (at the cost of reduced parallelism from unique RenderWorld borrows).
* Enables inserting into the same resource across multiple extract systems
* Enables using past RenderWorld state to inform future extract state (this should generally be avoided)
Ultimately this is just a subset of the functionality we want. In the future, it would be great to have "multi-world schedules" to enable fine grained parallelism on the render world during the extract step. But that is a research project that almost certainly won't make it into 0.6. This is a good interim solution that should easily port over to multi-world schedules if/when they land.
# Objective
- Remove all the `.system()` possible.
- Check for remaining missing cases.
## Solution
- Remove all `.system()`, fix compile errors
- 32 calls to `.system()` remains, mostly internals, the few others should be removed after #2446
# Objective
While looking at the code of `World`, I noticed two basic functions (`get` and `get_mut`) that are probably called a lot and with simple code that are not `inline`
## Solution
- Add benchmark to check impact
- Add `#[inline]`
```
group this pr main
----- ---- ----
world_entity/50000_entities 1.00 115.9±11.90µs ? ?/sec 1.71 198.5±29.54µs ? ?/sec
world_get/50000_entities_SparseSet 1.00 409.9±46.96µs ? ?/sec 1.18 483.5±36.41µs ? ?/sec
world_get/50000_entities_Table 1.00 391.3±29.83µs ? ?/sec 1.16 455.6±57.85µs ? ?/sec
world_query_for_each/50000_entities_SparseSet 1.02 121.3±18.36µs ? ?/sec 1.00 119.4±13.88µs ? ?/sec
world_query_for_each/50000_entities_Table 1.03 13.8±0.96µs ? ?/sec 1.00 13.3±0.54µs ? ?/sec
world_query_get/50000_entities_SparseSet 1.00 666.9±54.36µs ? ?/sec 1.03 687.1±57.77µs ? ?/sec
world_query_get/50000_entities_Table 1.01 584.4±55.12µs ? ?/sec 1.00 576.3±36.13µs ? ?/sec
world_query_iter/50000_entities_SparseSet 1.01 169.7±19.50µs ? ?/sec 1.00 168.6±32.56µs ? ?/sec
world_query_iter/50000_entities_Table 1.00 26.2±1.38µs ? ?/sec 1.91 50.0±4.40µs ? ?/sec
```
I didn't add benchmarks for the mutable path but I don't see how it could hurt to make it inline too...
