2
0
Fork 0
mirror of https://github.com/bevyengine/bevy synced 2025-02-15 21:48:34 +00:00
Commit graph

2431 commits

Author SHA1 Message Date
KDecay
46ff9fef98 Update input.rs docs in bevy_input ()
# Objective

- Part of the splitting process of .

## Solution

- Document `input.rs` inside of `bevy_input`.
2022-04-25 13:16:21 +00:00
Ferenc Tamás
2a5202637f bevy_utils: remove hardcoded log level limit ()
# Objective

- Debug logs are useful in release builds, but `tracing` logs are hard-capped (`release_max_level_info`) at the `info` level by `bevy_utils`.

## Solution

- This PR simply removes the limit in `bevy_utils` with no further actions.
- If any out-of-the box performance regressions arise, the steps to enable this `tracing` feature should be documented in a user guide in the future.

This PR closes  and closes .

## Alternatives considered

- Instruct the user to build with `debug-assertions` enabled: this is just a workaround, as it obviously enables all `debug-assertions` that affect more than logging itself.
- Re-exporting the feature from `tracing` and enabling it by default: I believe it just adds complexity and confusion, the `tracing` feature can also be re-enabled with one line in userland.

---

## Changelog

### Fixed

- Log level is not hard capped at `info` for release builds anymore.

## Migration Guide

- Maximum log levels for release builds is not enforced by Bevy anymore, to omit "debug" and "trace" level logs entirely from release builds, `tracing` must be added as a dependency with its `release_max_level_info` feature enabled in `Cargo.toml`. (`tracing = { version = "0.1", features = ["release_max_level_info"] }`)
2022-04-25 12:55:03 +00:00
Martin Dickopp
93cee3b3c9 Make Time::update_with_instant public for use in tests ()
# Objective

To test systems that implement frame rate-independent update logic, one needs to be able to mock `Time`. By mocking time, it's possible to write tests that confirm systems are frame rate-independent.

This is a follow-up PR to  by @ostwilkens and based on his work.

## Solution

To mock `Time`, one needs to be able to manually update the Time resource with an `Instant` defined by the developer. This can be achieved by making the existing `Time::update_with_instant` method public for use in tests. 

## Changelog

- Make `Time::update_with_instant` public
- Add doc to `Time::update_with_instant` clarifying that the method should not be called outside of tests.
- Add doc test to `Time` demonstrating how to use `update_with_instant` in tests.


Co-authored-by: Martin Dickopp <martin@zero-based.org>
2022-04-24 23:15:27 +00:00
François
1cd17e903f document the single threaded wasm task pool ()
# Objective

- The single threaded task pool is not documented
- This doesn't warn in CI as it's feature gated for wasm, but I'm tired of seeing the warnings when building in wasm

## Solution

- Document it
2022-04-24 22:57:04 +00:00
KDecay
00c6acc0cc Rename ElementState to ButtonState ()
# Objective

- Part of the splitting process of .

## Solution

- Rename `ElementState` to `ButtonState`

## Reasons

- The old name was too generic.
- If something can be pressed it is automatically button-like (thanks to @alice-i-cecile for bringing it up in ).
- The reason it is called `ElementState` is because that's how `winit` calls it.
- It is used to define if a keyboard or mouse **button** is pressed or not.
- Discussion in .

## Changelog

### Changed

- The `ElementState` type received a rename and is now called `ButtonState`.

## Migration Guide

- The `ElementState` type received a rename and is now called `ButtonState`. To migrate you just have to change every occurrence of `ElementState` to `ButtonState`.

Co-authored-by: KDecay <KDecayMusic@protonmail.com>
2022-04-24 22:57:02 +00:00
SpecificProtagonist
46acb7753c SystemSet::before and after: take AsSystemLabel ()
# Objective

`AsSystemLabel` has been introduced on system descriptors to make ordering systems more convenient, but `SystemSet::before` and `SystemSet::after` still take `SystemLabels` directly:

    use bevy::ecs::system::AsSystemLabel;
    /*…*/ SystemSet::new().before(foo.as_system_label()) /*…*/

is currently necessary instead of

    /*…*/ SystemSet::new().before(foo) /*…*/

## Solution

Use `AsSystemLabel` for `SystemSet`
2022-04-23 06:03:50 +00:00
TheRawMeatball
0fdb45ce90 Remove EntityMut::get_unchecked ()
The only way to soundly use this API is already encapsulated within `EntityMut::get`, so this api is removed.

# Migration guide

Replace calls to `EntityMut::get_unchecked` with calls to `EntityMut::get`.
2022-04-22 20:06:41 +00:00
CGMossa
7a0f46c21b fixes complaints about missing docs ()
# Objective

When using `derive(WorldQuery)`, then clippy complains with the following:

```rust
warning: missing documentation for a struct
  --> src\wild_boar_type\marker_vital_status.rs:35:17
   |
35 | #[derive(Debug, WorldQuery)]
   |                 ^^^^^^^^^^
   |
   = note: this warning originates in the derive macro `WorldQuery` (in Nightly builds, run with -Z macro-backtrace for more info)
```

## Solution

* Either `#[doc(hidden)]` or
* Add a generic documentation line to it.

I don't know what is preferred, but I'd gladly add it in here.
2022-04-22 08:45:04 +00:00
François
18c6a7b40e do not impl Component for Task ()
# Objective

- `Task` are `Component`.
- They should not.

## Solution

- Remove the impl, and update the example to show a wrapper.

 for reference
2022-04-22 06:29:38 +00:00
bjorn3
26c3b20f1c Remove some unused dependencies ()
This a commit I think would be uncontroversial that has been cherry-picked from https://github.com/bevyengine/bevy/pull/3886
2022-04-20 11:46:34 +00:00
bjorn3
e65f28d8d7 Remove parking_lot dependency from bevy_ecs ()
It is only used in some tests so any potential performance regressions don't matter.
2022-04-20 11:26:38 +00:00
KDecay
46321930f2 Update system.rs docs in bevy_input ()
# Objective

- Part of the splitting process of .

## Solution

- Document `system.rs` inside of `bevy_input`.
2022-04-20 03:06:31 +00:00
KDecay
7557f4db83 Update axis.rs docs in bevy_input ()
# Objective

- Part of the splitting process of .

## Solution

- Document `axis.rs` inside of `bevy_input`.
2022-04-18 21:26:54 +00:00
KDecay
b3e39d0a19 Update touch.rs docs ()
# Objective

- Part of the splitting process of .

## Solution

- Document `touch.rs` inside of `bevy_input`.
2022-04-18 18:25:44 +00:00
Hennadii Chernyshchyk
06d709b178 Make MaterialPipelineKey<T> fields public ()
# Objective

Fixes . This comment provides a very good explanation: https://github.com/bevyengine/bevy/issues/4507#issuecomment-1100905685.

## Solution

Make `MaterialPipelineKey` fields public.
2022-04-18 10:11:14 +00:00
Yutao Yuan
8d67832dfa Bump Bevy to 0.8.0-dev ()
# Objective

We should bump our version to 0.8.0-dev after releasing 0.7.0, according to our release checklist.

## Solution

Do it.
2022-04-17 23:04:52 +00:00
Charles
afbce46ade improve Commands doc comment ()
# Objective

- The current API docs of `Commands` is very short and is very opaque to newcomers.

## Solution

- Try to explain what it is without requiring knowledge of other parts of `bevy_ecs` like `World` or `SystemParam`.


Co-authored-by: Charles <IceSentry@users.noreply.github.com>
2022-04-17 18:17:49 +00:00
Daniel McNab
639fec20d6 Remove .system() ()
Free at last!

# Objective

- Using `.system()` is no longer needed anywhere, and anyone using it will have already gotten a deprecation warning.
- https://github.com/bevyengine/bevy/pull/3302 was a super special case for `.system()`, since it was so prevelant. However, that's no reason.
- Despite it being deprecated, another couple of uses of it have already landed, including in the deprecating PR.
   - These have all been because of doc examples having warnings not breaking CI - 🎟️?

## Solution

- Remove it.
- It's gone

---

## Changelog

- You can no longer use `.system()`

## Migration Guide

- You can no longer use `.system()`. It was deprecated in 0.7.0, and you should have followed the deprecation warning then. You can just remove the method call.

![image](https://user-images.githubusercontent.com/36049421/163688197-3e774a04-6f8f-40a6-b7a4-1330e0b7acf0.png)

- Thanks to the @TheRawMeatball  for producing
2022-04-16 21:33:51 +00:00
Carter Anderson
83c6ffb73c release 0.7.0 () 2022-04-15 18:05:37 +00:00
Robert Swain
b809e8e931 bevy_pbr: Fix ClusterConfig::None ()
# Objective

- Fix `ClusterConfig::None`
- This fix is from @robtfm but they didn't have time to submit it, so I am.

## Solution

- Always clear clusters and skip processing when `ClusterConfig::None`
- Conditionally remove `VisiblePointLights` from the view if it is present
2022-04-15 16:05:59 +00:00
Daniel McNab
424d4d26f1 A hack to work around minimising still being broken ()
# Objective

- https://github.com/bevyengine/bevy/pull/4098 still hasn't fixed minimisation on Windows.
- `Clusters.lights` is assumed to have the number of items given by the product of `Clusters.dimensions`'s axes.

## Solution

- Make that true in `clear`.
2022-04-15 11:22:48 +00:00
Dusty DeWeese
0d2b527faf Avoid windows with a physical size of zero ()
# Objective

Fix 

## Solution

Return `None` from `RenderTarget::get_physical_size` if either dimension is zero.
2022-04-15 07:32:21 +00:00
Hennadii Chernyshchyk
3b81a50a1a Fix crash in headless mode ()
# Objective

Fixes .

## Solution

Check if `RenderDevice` exists and add CI validation.
2022-04-15 07:13:37 +00:00
Robert Swain
c2a9d5843d Faster assign lights to clusters ()
# Objective

- Fixes 
- Fixes  
- Built on top of 
- Improve performance of `assign_lights_to_clusters`

## Solution

- Remove the OBB-based cluster light assignment algorithm and calculation of view space AABBs
- Implement the 'iterative sphere refinement' algorithm used in Just Cause 3 by Emil Persson as documented in the Siggraph 2015 Practical Clustered Shading talk by Persson, on pages 42-44 http://newq.net/dl/pub/s2015_practical.pdf
- Adapt to also support orthographic projections
- Add `many_lights -- orthographic` for testing many lights using an orthographic projection

## Results

- `assign_lights_to_clusters` in `many_lights` before this PR on an M1 Max over 1500 frames had a median execution time of 1.71ms. With this PR it is 1.51ms, a reduction of 0.2ms or 11.7% for this system.

---

## Changelog

- Changed: Improved cluster light assignment performance

Co-authored-by: robtfm <50659922+robtfm@users.noreply.github.com>
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-15 02:53:20 +00:00
François
d37cde8f1a fix feature location from ()
# Objective

- in , a feature for tracing was added to bevy_transform
- usage of that feature was moved to bevy_hierarchy, but the feature was not updated

## Solution

- add the feature to bevy_hierarchy, remove it from bevy_transform
2022-04-14 21:16:03 +00:00
François
8630b194dc add more logs when despawning entities ()
# Objective

- Provide more information when despawning an entity

## Solution

- Add a debug log when despawning an entity
- Add spans to the recursive ways of despawning an entity

```sh
RUST_LOG=debug cargo run --example panic --features trace
# RUST_LOG=debug needed to show debug logs from bevy_ecs
# --features trace needed to have the extra spans
...

DEBUG bevy_app:frame:stage{name=Update}:system_commands{name="panic::despawn_parent"}:command{name="DespawnRecursive" entity=0v0}: bevy_ecs::world: Despawning entity 1v0
DEBUG bevy_app:frame:stage{name=Update}:system_commands{name="panic::despawn_parent"}:command{name="DespawnRecursive" entity=0v0}: bevy_ecs::world: Despawning entity 0v0
```
2022-04-13 23:35:28 +00:00
Christian Hughes
16133de8cd WorldQuery derive macro now respects visibility ()
## Objective

Fixes .

## Solution

Inherit the visibility of the struct being derived for the `xxItem`, `xxFetch`, `xxState` structs.

Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-13 21:50:45 +00:00
Martin Dickopp
995f85e157 Fix typos in Time::last_update and Time::time_since_startup documentation ()
# Objective

The documentation of the `Time::last_update` and `Time::time_since_startup` methods contains typos. It uses apostrophe instead of backtick characters around `Instant` and `Duration`, so that these words are not recognized as identifiers in the generated API documentation. This should be fixed.

## Solution

Fix the typos.
2022-04-13 21:34:21 +00:00
Johannes Hackel
621f3969c7 Switch render order to 2d-on-top-of-3d ()
Because  got merged only switching the render order is left to fix 
2022-04-13 20:05:34 +00:00
Carter Anderson
8783fae7de Use "many" instead of "multiple" consistently ()
We missed a couple of these renames in 
2022-04-13 00:35:47 +00:00
Dusty DeWeese
5a297d7903 Reuse texture when resolving multiple passes ()
# Objective

Fixes https://github.com/bevyengine/bevy/issues/3499

## Solution

Uses a `HashMap` from `RenderTarget` to sampled textures when preparing `ViewTarget`s to ensure that two passes with the same render target get sampled to the same texture.

This builds on and depends on https://github.com/bevyengine/bevy/pull/3412, so this will be a draft PR until  is merged. All changes for this PR are in the last commit.
2022-04-12 19:27:30 +00:00
Jakob Hellermann
193e8c4ada scene_viewer: load cameras ()
# Objective

glTF files can contain cameras. Currently the scene viewer example uses _a_ camera defined in the file if possible, otherwise it spawns a new one. It would be nice if instead it could load all the cameras and cycle through them, while also having a separate user-controller camera.

## Solution

- instead of just a camera that is already defined, always spawn a new separate user-controller camera
- maintain a list of loaded cameras and cycle through them (wrapping to the user-controller camera) when pressing `C`

This matches the behavious that https://github.khronos.org/glTF-Sample-Viewer-Release/ has.

## Implementation notes

- The gltf scene asset loader just spawns the cameras into the world, but does not return a mapping of camera index to bevy entity. So instead the scene_viewer example just collects all spawned cameras with a good old `query.iter().collect()`, so the order is unspecified and may change between runs.

## Demo

https://user-images.githubusercontent.com/22177966/161826637-40161482-5b3b-4df5-aae8-1d5e9b918393.mp4


using the virtual city glTF sample file: https://github.com/KhronosGroup/glTF-Sample-Models/tree/master/2.0/VC

Co-authored-by: Jakob Hellermann <hellermann@sipgate.de>
2022-04-11 22:56:06 +00:00
François
e8cd2fc727 Audio control at start of playback ()
# Objective

- While playing with volume, I noticed that when setting the volume just after playback start, I still get a few milliseconds at normal volume

## Solution

- Replace `play_in_loop` with `play_with_settings` that allows from more controls
- Adds a `PlaybackSettings` to specify the settings from start. Can be used: `PlaybackSettings::LOOP.with_volume(0.75)`
2022-04-08 23:12:30 +00:00
Jakob Hellermann
c12ee81822 bevy_app: add tracing event with tracy.frame_mark ()
Currently `tracy` interprets the entire trace as one frame because the marker for frames isn't being recorded.

~~When an event with `tracy.trace_marker=true` is recorded, `tracing-tracy` will mark the frame as finished:
<aa0b96b2ae/tracing-tracy/src/lib.rs (L240)>~~

~~Unfortunately this leads to~~
```rs
INFO bevy_app:frame: bevy_app::app: finished frame tracy.frame_mark=true
```
~~being printed every frame (we can't use DEBUG because bevy_log sets `max_release_level_info`.~~

Instead of emitting an event that gets logged every frame, we can depend on tracy-client itself and call `finish_continuous_frame!();`
2022-04-08 22:50:23 +00:00
Kirillov Kirill
3756181e23 Change scaling mode to FixedHorizontal ()
# Objective

- Fixes the issue with orthographic camera imported from glTF not displaying anything (mentioned in ).

## Solution

- This was due to wrong scaling mode being used. This PR simply changes WindowSize scaling mode to FixedHorizontal.

## Important Note

Currently, othographic scale in Blender, three.js, and possibly other software does not translate to Bevy (via glTF) because their developers have [misinterpreted the spec](https://github.com/KhronosGroup/glTF/issues/1663#issuecomment-618194015). The camera parameters have been clarified in glTF 2.0, which was released on October of 2021. In Blender 3.0.1 this issue has **not** been fixed yet. If you are importing orthographic cameras from Blender, you have to divide the scale by 2.
2022-04-08 17:28:32 +00:00
harudagondi
64d217823d Allow iter combinations on queries with filters ()
# Objective

- Previously, `iter_combinations()` does not work on queries that have filters.
- Fixes 

## Solution

- Derived Copy on all `*Fetch<T>` structs, and manually implemented `Clone` to allow the test to pass (`.count()` does not work on `QueryCombinationIter` when `Clone` is derived)


Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-08 00:21:24 +00:00
Kirillov Kirill
01bdf67c33 Improve the set_active_camera system ()
# Objective

- Make `set_active_camera` system correctly respond to camera deletion, while preserving its correct behavior on first ever frame and any consequent frame, and with multiple cameras of the same type available in the world.
- Fixes 

## Solution

- Add a check that the entity referred to by `ActiveCamera` still exists in the world.
2022-04-07 23:30:47 +00:00
François
d478723e19 insert the gltf mesh name on the entity if there is one ()
# Objective

- In glTF, mesh can be named. This named is used to be able to reference the mesh, but not as a component on the entity
- Bevy only added the node name to the parent node.


## Solution

- Also adds the name on the mesh entity if there is one.

Limitation: In glTF, it's possible to have one mesh (which can be named) corresponding to several primitives (which can't, but are the actual mesh). I added the mesh name to the entity with the `PbrBundle` matching the primitives, which means that a mesh with several primitives would all have the same name. I think this is acceptable...
2022-04-07 22:19:32 +00:00
François
4feb0d520a increase the maximum number of point lights with shadows to the max supported by the device ()
# Objective

- Being limited to 10 pointlights with shadow is very limiting

## Solution

- Raise the limit
2022-04-07 21:55:31 +00:00
François
8268e7fa9e expose extras from gltf nodes ()
fixes  

expose the `extras` field value as a string
2022-04-07 21:30:52 +00:00
Torne Wuff
b1afe2dcca Make System responsible for updating its own archetypes ()
# Objective

- Make it possible to use `System`s outside of the scheduler/executor without having to define logic to track new archetypes and call `System::add_archetype()` for each.

## Solution

- Replace `System::add_archetype(&Archetype)` with `System::update_archetypes(&World)`, making systems responsible for tracking their own most recent archetype generation the way that `SystemState` already does.

This has minimal (or simplifying) effect on most of the code with the exception of `FunctionSystem`, which must now track the latest `ArchetypeGeneration` it saw instead of relying on the executor to do it.

Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-07 20:50:43 +00:00
Gabriel Bourgeois
6f16580b8a Fix clicked UI nodes getting reset when hovering child nodes ()
# Objective

Fixes 

## Solution

When resetting a node's `Interaction` to `None`, ignore any `Clicked` node because that should be handled by the mouse release check exclusively.
2022-04-07 19:08:09 +00:00
Daniel McNab
21a875d67b Some small changes related to run criteria piping ()
Remove the 'chaining' api, as it's peculiar

~~Implement the label traits for `Box<dyn ThatTrait>` (n.b. I'm not confident about this change, but it was the quickest path to not regressing)~~

Remove the need for '`.system`' when using run criteria piping
2022-04-07 19:08:08 +00:00
François
9d54f33974 Skinned extraction speedup ()
# Objective

- While animating 501 https://github.com/KhronosGroup/glTF-Sample-Models/tree/master/2.0/BrainStem, I noticed things were getting a little slow
- Looking in tracy, the system `extract_skinned_meshes` is taking a lot of time, with a mean duration of 15.17ms

## Solution

- ~~Use `Vec` instead of a `SmallVec`~~
- ~~Don't use an temporary variable~~
- Compute the affine matrix as an `Affine3A` instead
- Remove the `temp` vec

| |mean|
|---|---|
|base|15.17ms|
|~~vec~~|~~9.31ms~~|
|~~no temp variable~~|~~11.31ms~~|
|removing the temp vector|8.43ms|
|affine|13.21ms|
|all together|7.23ms|
2022-04-07 16:16:36 +00:00
Robert Swain
c5963b4fd5 Use storage buffers for clustered forward point lights ()
# Objective

- Make use of storage buffers, where they are available, for clustered forward bindings to support far more point lights in a scene
- Fixes  
- Based on top of  

This branch on an M1 Max can keep 60fps with about 2150 point lights of radius 1m in the Sponza scene where I've been testing. The bottleneck is mostly assigning lights to clusters which grows faster than linearly (I think 1000 lights was about 1.5ms and 5000 was 7.5ms). I have seen papers and presentations leveraging compute shaders that can get this up to over 1 million. That said, I think any further optimisations should probably be done in a separate PR.

## Solution

- Add `RenderDevice` to the `Material` and `SpecializedMaterial` trait `::key()` functions to allow setting flags on the keys depending on feature/limit availability
- Make `GpuPointLights` and `ViewClusterBuffers` into enums containing `UniformVec` and `StorageBuffer` variants. Implement the necessary API on them to make usage the same for both cases, and the only difference is at initialisation time.
- Appropriate shader defs in the shader code to handle the two cases

## Context on some decisions / open questions

- I'm using `max_storage_buffers_per_shader_stage >= 3` as a check to see if storage buffers are supported. I was thinking about diving into 'binding resource management' but it feels like we don't have enough use cases to understand the problem yet, and it is mostly a separate concern to this PR, so I think it should be handled separately.
- Should `ViewClusterBuffers` and `ViewClusterBindings` be merged, duplicating the count variables into the enum variants?


Co-authored-by: Carter Anderson <mcanders1@gmail.com>
2022-04-07 16:16:35 +00:00
Robert Swain
579928e8e0 bevy_pbr: Support flipping tangent space normal map y for DirectX normal maps ()
# Objective

- Normal maps authored for DirectX use a left-handed convention and have their tangent space normal in the texture inverted from what we need. Support this.
- Details here: https://doc.babylonjs.com/divingDeeper/materials/advanced/normalMaps

## Solution

- Add a `StandardMaterial` `flip_normal_map_y` boolean field
- Add a `STANDARDMATERIAL_FLIP_NORMAL_MAP_Y` flag to `StandardMaterialFlags` and in the PBR shader
- Flip the y-component of the tangent space normal just after sampling it from the normal map texture

## Screenshots

### Before

<img width="1392" alt="Screenshot 2022-04-06 at 21 04 44" src="https://user-images.githubusercontent.com/302146/162050314-e7bfaaf6-9ee1-4756-9821-f6f5ff78f508.png">

### After

<img width="1392" alt="Screenshot 2022-04-06 at 21 03 39" src="https://user-images.githubusercontent.com/302146/162050255-36ee0745-1d79-4fd2-9a1c-18085376b643.png">

---

## Changelog

- Added: Support for flipping the normal map texture y component for normal maps authored for use with DirectX
2022-04-07 15:50:14 +00:00
KDecay
f907d67d7e Fix formatting, spelling, phrasing and consistency ()
# Objective

The `bevy_app` crate has a lot of inconsistencies in the documentation (formatting, spelling, phrasing, consistency).

## Solution

Make it more consistent.
2022-04-05 22:36:02 +00:00
François
30878e3a7d add AnimationPlayer component only on scene roots that are also animation roots ()
# Objective

- Fix  
- The scene has two root nodes, with the second one being the animation root

## Solution

- Check all scene root nodes, and add the `AnimationPlayer` component to nodes that are also animation roots
2022-04-05 21:52:00 +00:00
TheRawMeatball
73edb11db6 Add more FromWorld implementations ()
# Objective

Make `FromWorld` more useful for abstractions with a form similar to
```rs
trait FancyAbstraction {
  type PreInitializedData: FromWorld;
}
```

## Solution

Add a `FromWorld` implementation for `SystemState` as well as a way to group together multiple `FromWorld` implementing types as one.

Note: I plan to follow up this PR with another to add `Local` support to exclusive systems, which should get a fair amount of use from the `FromWorld` implementation on `SystemState`.
2022-04-05 20:04:34 +00:00
Hennadii Chernyshchyk
ea6e6f7db2 Do not crash if RenderDevice doesn't exist ()
# Objective

Avoid crashing if `RenderDevice` doesn't exist (required for headless mode).
Fixes .

## Solution

Use `CompressedImageFormats::all()` if there is no `RenderDevice`.
2022-04-05 19:37:23 +00:00