Huge thanks to @maniwani, @devil-ira, @hymm, @cart, @superdump and @jakobhellermann for the help with this PR.
# Objective
- Followup #6587.
- Minimal integration for the Stageless Scheduling RFC: https://github.com/bevyengine/rfcs/pull/45
## Solution
- [x] Remove old scheduling module
- [x] Migrate new methods to no longer use extension methods
- [x] Fix compiler errors
- [x] Fix benchmarks
- [x] Fix examples
- [x] Fix docs
- [x] Fix tests
## Changelog
### Added
- a large number of methods on `App` to work with schedules ergonomically
- the `CoreSchedule` enum
- `App::add_extract_system` via the `RenderingAppExtension` trait extension method
- the private `prepare_view_uniforms` system now has a public system set for scheduling purposes, called `ViewSet::PrepareUniforms`
### Removed
- stages, and all code that mentions stages
- states have been dramatically simplified, and no longer use a stack
- `RunCriteriaLabel`
- `AsSystemLabel` trait
- `on_hierarchy_reports_enabled` run criteria (now just uses an ad hoc resource checking run condition)
- systems in `RenderSet/Stage::Extract` no longer warn when they do not read data from the main world
- `RunCriteriaLabel`
- `transform_propagate_system_set`: this was a nonstandard pattern that didn't actually provide enough control. The systems are already `pub`: the docs have been updated to ensure that the third-party usage is clear.
### Changed
- `System::default_labels` is now `System::default_system_sets`.
- `App::add_default_labels` is now `App::add_default_sets`
- `CoreStage` and `StartupStage` enums are now `CoreSet` and `StartupSet`
- `App::add_system_set` was renamed to `App::add_systems`
- The `StartupSchedule` label is now defined as part of the `CoreSchedules` enum
- `.label(SystemLabel)` is now referred to as `.in_set(SystemSet)`
- `SystemLabel` trait was replaced by `SystemSet`
- `SystemTypeIdLabel<T>` was replaced by `SystemSetType<T>`
- The `ReportHierarchyIssue` resource now has a public constructor (`new`), and implements `PartialEq`
- Fixed time steps now use a schedule (`CoreSchedule::FixedTimeStep`) rather than a run criteria.
- Adding rendering extraction systems now panics rather than silently failing if no subapp with the `RenderApp` label is found.
- the `calculate_bounds` system, with the `CalculateBounds` label, is now in `CoreSet::Update`, rather than in `CoreSet::PostUpdate` before commands are applied.
- `SceneSpawnerSystem` now runs under `CoreSet::Update`, rather than `CoreStage::PreUpdate.at_end()`.
- `bevy_pbr::add_clusters` is no longer an exclusive system
- the top level `bevy_ecs::schedule` module was replaced with `bevy_ecs::scheduling`
- `tick_global_task_pools_on_main_thread` is no longer run as an exclusive system. Instead, it has been replaced by `tick_global_task_pools`, which uses a `NonSend` resource to force running on the main thread.
## Migration Guide
- Calls to `.label(MyLabel)` should be replaced with `.in_set(MySet)`
- Stages have been removed. Replace these with system sets, and then add command flushes using the `apply_system_buffers` exclusive system where needed.
- The `CoreStage`, `StartupStage, `RenderStage` and `AssetStage` enums have been replaced with `CoreSet`, `StartupSet, `RenderSet` and `AssetSet`. The same scheduling guarantees have been preserved.
- Systems are no longer added to `CoreSet::Update` by default. Add systems manually if this behavior is needed, although you should consider adding your game logic systems to `CoreSchedule::FixedTimestep` instead for more reliable framerate-independent behavior.
- Similarly, startup systems are no longer part of `StartupSet::Startup` by default. In most cases, this won't matter to you.
- For example, `add_system_to_stage(CoreStage::PostUpdate, my_system)` should be replaced with
- `add_system(my_system.in_set(CoreSet::PostUpdate)`
- When testing systems or otherwise running them in a headless fashion, simply construct and run a schedule using `Schedule::new()` and `World::run_schedule` rather than constructing stages
- Run criteria have been renamed to run conditions. These can now be combined with each other and with states.
- Looping run criteria and state stacks have been removed. Use an exclusive system that runs a schedule if you need this level of control over system control flow.
- For app-level control flow over which schedules get run when (such as for rollback networking), create your own schedule and insert it under the `CoreSchedule::Outer` label.
- Fixed timesteps are now evaluated in a schedule, rather than controlled via run criteria. The `run_fixed_timestep` system runs this schedule between `CoreSet::First` and `CoreSet::PreUpdate` by default.
- Command flush points introduced by `AssetStage` have been removed. If you were relying on these, add them back manually.
- Adding extract systems is now typically done directly on the main app. Make sure the `RenderingAppExtension` trait is in scope, then call `app.add_extract_system(my_system)`.
- the `calculate_bounds` system, with the `CalculateBounds` label, is now in `CoreSet::Update`, rather than in `CoreSet::PostUpdate` before commands are applied. You may need to order your movement systems to occur before this system in order to avoid system order ambiguities in culling behavior.
- the `RenderLabel` `AppLabel` was renamed to `RenderApp` for clarity
- `App::add_state` now takes 0 arguments: the starting state is set based on the `Default` impl.
- Instead of creating `SystemSet` containers for systems that run in stages, simply use `.on_enter::<State::Variant>()` or its `on_exit` or `on_update` siblings.
- `SystemLabel` derives should be replaced with `SystemSet`. You will also need to add the `Debug`, `PartialEq`, `Eq`, and `Hash` traits to satisfy the new trait bounds.
- `with_run_criteria` has been renamed to `run_if`. Run criteria have been renamed to run conditions for clarity, and should now simply return a bool.
- States have been dramatically simplified: there is no longer a "state stack". To queue a transition to the next state, call `NextState::set`
## TODO
- [x] remove dead methods on App and World
- [x] add `App::add_system_to_schedule` and `App::add_systems_to_schedule`
- [x] avoid adding the default system set at inappropriate times
- [x] remove any accidental cycles in the default plugins schedule
- [x] migrate benchmarks
- [x] expose explicit labels for the built-in command flush points
- [x] migrate engine code
- [x] remove all mentions of stages from the docs
- [x] verify docs for States
- [x] fix uses of exclusive systems that use .end / .at_start / .before_commands
- [x] migrate RenderStage and AssetStage
- [x] migrate examples
- [x] ensure that transform propagation is exported in a sufficiently public way (the systems are already pub)
- [x] ensure that on_enter schedules are run at least once before the main app
- [x] re-enable opt-in to execution order ambiguities
- [x] revert change to `update_bounds` to ensure it runs in `PostUpdate`
- [x] test all examples
- [x] unbreak directional lights
- [x] unbreak shadows (see 3d_scene, 3d_shape, lighting, transparaency_3d examples)
- [x] game menu example shows loading screen and menu simultaneously
- [x] display settings menu is a blank screen
- [x] `without_winit` example panics
- [x] ensure all tests pass
- [x] SubApp doc test fails
- [x] runs_spawn_local tasks fails
- [x] [Fix panic_when_hierachy_cycle test hanging](https://github.com/alice-i-cecile/bevy/pull/120)
## Points of Difficulty and Controversy
**Reviewers, please give feedback on these and look closely**
1. Default sets, from the RFC, have been removed. These added a tremendous amount of implicit complexity and result in hard to debug scheduling errors. They're going to be tackled in the form of "base sets" by @cart in a followup.
2. The outer schedule controls which schedule is run when `App::update` is called.
3. I implemented `Label for `Box<dyn Label>` for our label types. This enables us to store schedule labels in concrete form, and then later run them. I ran into the same set of problems when working with one-shot systems. We've previously investigated this pattern in depth, and it does not appear to lead to extra indirection with nested boxes.
4. `SubApp::update` simply runs the default schedule once. This sucks, but this whole API is incomplete and this was the minimal changeset.
5. `time_system` and `tick_global_task_pools_on_main_thread` no longer use exclusive systems to attempt to force scheduling order
6. Implemetnation strategy for fixed timesteps
7. `AssetStage` was migrated to `AssetSet` without reintroducing command flush points. These did not appear to be used, and it's nice to remove these bottlenecks.
8. Migration of `bevy_render/lib.rs` and pipelined rendering. The logic here is unusually tricky, as we have complex scheduling requirements.
## Future Work (ideally before 0.10)
- Rename schedule_v3 module to schedule or scheduling
- Add a derive macro to states, and likely a `EnumIter` trait of some form
- Figure out what exactly to do with the "systems added should basically work by default" problem
- Improve ergonomics for working with fixed timesteps and states
- Polish FixedTime API to match Time
- Rebase and merge #7415
- Resolve all internal ambiguities (blocked on better tools, especially #7442)
- Add "base sets" to replace the removed default sets.
# Objective
- Avoid hitting the 6 hours default timeout
- Waiting for 6 hours for a job to fail is wasteful and slow down CI for other PRs
## Solution
- Put shorter timeouts on all jobs
# Objective
- Make CI friendlier
## Solution
- CI now says hello to new contributor
- for some jobs with non obvious solutions to failures, give more context
- example run should say which example failed
- example doc should say the next action to do (add metadata or run the update script)
- MSRV will say when it needs updating
I'm not completely sure everything is working and will try to trigger failures in this PR
# Objective
- Fixes#6777, fixes#2998, replaces #5518
- Help avoid confusing error message when using an older version of Rust
## Solution
- Add the `rust-version` field to `Cargo.toml`
- Add a CI job checking the MSRV
- Add the job to bors
# Objective
There isn't really a way to test that code using bevy_reflect compiles or doesn't compile for certain scenarios. This would be especially useful for macro-centric PRs like #6511 and #6042.
## Solution
Using `bevy_ecs_compile_fail_tests` as reference, added the `bevy_reflect_compile_fail_tests` crate.
Currently, this crate contains a very simple test case. This is so that we can get the basic foundation of this crate agreed upon and merged so that more tests can be added by other PRs.
### Open Questions
- [x] Should this be added to CI? (Answer: Yes)
---
## Changelog
- Added the `bevy_reflect_compile_fail_tests` crate for testing compilation errors
# Objective
- run examples is failing with `xvfb-run: error: Xvfb failed to start`
## Solution
- rollback ubuntu version for run-examples to 20.04. latest is 22.04
## Notes
- this is just a quick fix and someone should probably get it working on 22.04. I'll make an issue for that if this gets merged.
# Objective
- Fix CI issue with updated `cargo-app`
## Solution
- Move the Android example to its own package. It's not necessary for the CI fix, but it's cleaner, mimic the iOS example, and easier to reuse for someone wanting to setup android support in their project
- Build the package in CI instead of the example
The Android example is still working on my android device with this change 👍
# Objective
- Fix disabling features in bevy_ecs (broken by #5630)
- Add tests in CI for bevy_ecs, bevy_reflect and bevy as those crates could be use standalone
# Objective
I was about to submit a PR to add these two examples to `bevy-website` and re-discovered the inconsistency.
Although it's not a major issue on the website where only the filenames are shown, this would help to visually distinguish the two examples in the list because the names are very prominent.
This also helps out when fuzzy-searching the codebase for these files.
## Solution
Rename `shapes` to `2d_shapes`. Now the filename matches the example name, and the naming structure matches the 3d example.
## Notes
@Nilirad proposed this in https://github.com/bevyengine/bevy/pull/4613#discussion_r862455631 but it had slipped away from my brain at that time.
# Objective
Fixes#5668.
The Rust version used in the CI `build` step previously depended on the default Rust version defined by GitHub in the Ubuntu image: <https://github.com/actions/runner-images/blob/main/images/linux/Ubuntu2004-Readme.md#rust-tools>
This currently doesn't allow us to use Rust 1.63 features until this version is updated.
## Solution
We now use the `actions-rs/toolchain@v1` action to always use the latest stable Rust version.
This is already used for other CI jobs that we have.
# Objective
- Fixes#5463
- set ANDROID_NDK_ROOT
- GitHub recently updated their ubuntu container, removing some of the android environment variable: ca5d04c7da
- `cargo-apk` is not reading the new environment variable: 9a8be258a9/ndk-build/src/ndk.rs (L33-L38)
- this also means CI will now use the latest android NDK, I don't know if that's an issue
# Objective
- Running examples on windows crash due to full disk
- The prebuild step was not being reused and consuming extra space
## Solution
- Use the exact same command to prebuild to ensure it will be reused
- Also on linux
# Objective
Rust's nightly builds semi-regularly break us (or our dependencies). This creates churn and angst when we're just trying to get our jobs done.
We do still want nightly builds for a variety of reasons:
* cargo-udeps requires nightly and likely always will.
* Helps us catch rust nightly bugs quickly. We're "good citizens" if we regularly report regressions.
* Lets us prepare for "actual expected breakage" ahead of stable releases so we avoid breaking main users.
## Solution
* This pr parameterizes the nightly toolchain, making it an easy one-liner to pin our builds to a specific nightly, when required.
* Put nightly jobs to check wasm and nightly on their own matrix. I also removed tests on nightly linux, just build check
* alternative to https://github.com/bevyengine/bevy/pull/5329
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Reduce confusion as the example opens a window and isn't truly "headless"
- Fixes https://github.com/bevyengine/bevy/issues/5260.
## Solution
- Rename the example and add to the docs that the window is expected.
This gives users a hint to use the GitHub "Discussions" for questions about bevy, instead of filing an issue.
## Objective
> Users sometimes file unhelpful issues when asking questions about how to use Bevy.
We should provide a link to the Discussion board in the list of "new Issue" options. This hopefully allows users to better find this option and reduces the number of question-issues.
- fixes#5150
## Solution
- add a small config.yml that configures the link
Looks like this (currently live on my local fork https://github.com/themasch/bevy/issues/new/choose):
---
## Open questsions
- I am unsure about the wording.
# Objective
When `miri` runs in our build system to detect unsoundness, its output can be very unhelpful, as the tests are all run in parallel.
## Solution
Add a comment documenting the extremely obvious 10/10 command used by @BoxyUwU in #4959.
I've stuck this in the CI file, as it seems like the most obvious place to check when frustrated. I didn't put it in CONTRIBUTING.md because this is an eldritch abomination and will never be useful to new contributors.
# Objective
Fixes#5155. This *should* work now that the semver breaking dependency of the CI crate got yanked, but we'll see what CI has to say about it.
The first leak:
```rust
#[test]
fn blob_vec_drop_empty_capacity() {
let item_layout = Layout:🆕:<Foo>();
let drop = drop_ptr::<Foo>;
let _ = unsafe { BlobVec::new(item_layout, Some(drop), 0) };
}
```
this is because we allocate the swap scratch in blobvec regardless of what the capacity is, but we only deallocate if capacity is > 0
The second leak:
```rust
#[test]
fn panic_while_overwriting_component() {
let helper = DropTestHelper::new();
let res = panic::catch_unwind(|| {
let mut world = World::new();
world
.spawn()
.insert(helper.make_component(true, 0))
.insert(helper.make_component(false, 1));
println!("Done inserting! Dropping world...");
});
let drop_log = helper.finish(res);
assert_eq!(
&*drop_log,
[
DropLogItem::Create(0),
DropLogItem::Create(1),
DropLogItem::Drop(0),
]
);
}
```
this is caused by us not running the drop impl on the to-be-inserted component if the drop impl of the overwritten component panics
---
managed to figure out where the leaks were by using this 10/10 command
```
cargo --quiet test --lib -- --list | sed 's/: test$//' | MIRIFLAGS="-Zmiri-disable-isolation" xargs -n1 cargo miri test --lib -- --exact
```
which runs every test one by one rather than all at once which let miri actually tell me which test had the leak 🙃
# Objective
- Follow suggestion from https://github.com/bevyengine/bevy/pull/4984#issuecomment-1152949640
## Solution
- Unpin nightly, disable weak memory emulation
---
This failed the miri job in my branch with the following error:
```
error: Undefined Behavior: attempting a read access using <untagged> at alloc198028[0x0], but that tag does not exist in the borrow stack for this location
--> /home/runner/.cargo/registry/src/github.com-1ecc6299db9ec823/once_cell-1.12.0/src/imp_std.rs:177:28
|
177 | let next = (*waiter).next;
| ^^^^^^^^^^^^^^
| |
| attempting a read access using <untagged> at alloc198028[0x0], but that tag does not exist in the borrow stack for this location
| this error occurs as part of an access at alloc198028[0x0..0x8]
|
= help: this indicates a potential bug in the program: it performed an invalid operation, but the rules it violated are still experimental
= help: see https://github.com/rust-lang/unsafe-code-guidelines/blob/master/wip/stacked-borrows.md for further information
```
@BoxyUwU could you take a look? I guess it's related to the issue mentioned in https://github.com/rust-lang/miri/issues/2223
# Objective
- Make Bevy work on android
## Solution
- Update android metadata and add a few more
- Set the target sdk to 31 as it will soon (in august) be the minimum sdk level for play store
- Remove the custom code to create an activity and use ndk-glue macro instead
- Delay window creation event on android
- Set the example with compatibility settings for wgpu. Those are needed for Bevy to work on my 2019 android tablet
- Add a few details on how to debug in case of failures
- Fix running the example on emulator. This was failing because of the name of the example
Bevy still doesn't work on android with this, audio features need to be disabled because of an ndk-glue version mismatch: rodio depends on 0.6.2, winit on 0.5.2. You can test with:
```
cargo apk run --release --example android_example --no-default-features --features "bevy_winit,render"
```
# Objective
CI is now failing with some changes that landed in 1.62.
## Solution
* Fix an unused lifetime by using it (we double-used the `w` lifetime).
* Update compile_fail error messages
* temporarily disable check-unused-dependencies
# Objective
- Have information about examples only in one place that can be used for the repo and for the website (and remove the need to keep a list of example to build for wasm in the website 75acb73040/generate-wasm-examples/generate_wasm_examples.sh (L92-L99))
## Solution
- Add metadata about examples in `Cargo.toml`
- Build the `examples/README.md` from a template using those metadata. I used tera as the template engine to use the same tech as the website.
- Make CI fail if an example is missing metadata, or if the readme file needs to be updated (the command to update it is displayed in the failed step in CI)
## Remaining To Do
- After the next release with this merged in, the website will be able to be updated to use those metadata too
- I would like to build the examples in wasm and make them available at http://dev-docs.bevyengine.org/ but that will require more design
- https://github.com/bevyengine/bevy-website/issues/299 for other ToDos
Co-authored-by: Readme <github-actions@github.com>
# Objective
- Fix timeout in miri
## Solution
- Use a nightly version from before the issue happened: 2022-06-08
- To be checked after https://github.com/rust-lang/miri/issues/2223 is fixed
# Objective
- Run examples in WASM in CI
- Fix#4817
## Solution
- on feature `bevy_ci_testing`
- add an extra log message before exiting
- when building for wasm, read CI config file at compile time
- add a simple [playwright](https://playwright.dev) test script that opens the browser then waits for the success log, and takes a screenshot
- add a CI job that runs the playwright test for Chromium and Firefox on one example (lighting) and save the screenshots
- Firefox screenshot is good (with some clusters visible)
- Chromium screenshot is gray, I don't know why but it's logging `GPU stall due to ReadPixels`
- Webkit is not enabled for now, to revisit once https://bugs.webkit.org/show_bug.cgi?id=234926 is fixed or worked around
- the CI job only runs on bors validation
example run: https://github.com/mockersf/bevy/actions/runs/2361673465. The screenshots can be downloaded
# Objective
While playing with the code, I found some problems in the recently merged version-bumping workflow:
- Most importantly, now that we are using `0.8.0-dev` in development, the workflow will try to bump it to `0.9.0` 😭
- The crate filter is outdated now that we have more crates in `tools`.
- We are using `bevy@users.noreply.github.com`, but according to [Github help](https://docs.github.com/en/account-and-profile/setting-up-and-managing-your-personal-account-on-github/managing-email-preferences/setting-your-commit-email-address#about-commit-email-addresses), that email address means "old no-reply email format for the user `bevy`". It is currently not associated with any account, but I feel this is still not appropriate here.
## Solution
- Create a new workflow, `Post-release version bump`, that should be run after a release and bumps version from `0.X.0` to `0.X+1.0-dev`. Unfortunately, cargo-release doesn't have a builtin way to do this, so we need to parse and increment the version manually.
- Add the new crates in `tools` to exclusion list. Also removes the dependency version specifier from `bevy_ecs_compile_fail_tests`. It is not in the workspace so the dependency version will not get automatically updated by cargo-release.
- Change the author email to `41898282+github-actions[bot]@users.noreply.github.com`. According to the discussion [here](https://github.com/actions/checkout/issues/13#issuecomment-724415212) and [here](https://github.community/t/github-actions-bot-email-address/17204/6), this is the email address associated with the github-actions bot account.
- Also add the workflows to our release checklist.
See infmagic2047#5 and infmagic2047#6 for examples of release and post-release PRs.
# Objective
- Ensure future Bevy releases happens smoothly
## Solution
- Add a workflow that will open a PR updating all Bevy crate that can be created manually
example PR opened: https://github.com/mockersf/bevy/pull/62
The day from this PR does not need to be the release day, it will just open the PR to prepare it. Later if we feel confident, it could push automatically to crates.io.
how to trigger the workflow: https://docs.github.com/en/actions/managing-workflow-runs/manually-running-a-workflow
# Objective
- ~~Running examples on Linux in CI timeout~~Linux is back!
- But hey we can run examples on windows too!
## Solution
- Run examples on windows daily
- I also added a 30 minutes timeout so that when it explodes, it doesn't explodes in 6 hours (the default timeout)
- And simplified the linux examples by not requiring a custom feature set
# Objective
1. "What you expected to happen" and "what actually happened" often involves trivial duplication.
2. "Please provide full reproduction steps" is not helpful advice to new contributors.
3. The OS field was commonly useless or inadequate.
4. The description for "additional information" effectively just repeated the title of the field.
## Solution
1. Unify these fields into a single "what went wrong" field.
2. Provide an example of a useful reproduction.
3. Replace OS field with an optional "Setup Information" field that captures information about other critical setup like Rust version and hardware.
4. Provide helpful advice about what sort of information may be useful to add.
# Objective
This fails constantly and causes more pain than it is worth.
## Solution
Remove dead link checks.
Alternative to #4837, which is more granular but ironically still fails to build. I'm in favor of the nuclear option.
Fixes#4575
# Objective
- When Miri is failing, it can be very slow to do so
<img width="1397" alt="Screenshot 2022-05-14 at 03 05 40" src="https://user-images.githubusercontent.com/8672791/168405111-c5e27d63-7a5a-4a5e-b679-abbeeb3201d2.png">
## Solution
- Set the timeout for Miri to 60 minutes (it's 6 hours by default). It runs in around 10 minutes when successful
- Fix cache key as it was set to the same as another task that doesn't build with the same parameters
# Objective
- New PRs are labeled with Needs-Triage, but this is unhelpful and creates busy work: it's just as easy to check for unlabelled PRs, especially now that we no longer have an unlabelled backlog.
Note: this is not true for issues. Issues start with at least one label based on which template they use, and so there's no good way to filter for issues that need attention from the triage team.
## Solution
- Remove responsible CI tasks.
# Objective
- Original objective was to add doc build warning check to the ci local execution
- I somewhat deviated and changed other things...
## Solution
`cargo run -p ci` can now take more parameters:
* `format` - cargo fmt
* `clippy` - clippy
* `compile-fail` - bevy_ecs_compile_fail_tests tests
* `test` - tests but not doc tests and do not build examples
* `doc-test` - doc tests
* `doc-check` - doc build and warnings
* `bench-check` - check that benches build
* `example-check` - check that examples build
* `lints` - group - run lints and format and clippy
* `doc` - group - run doc-test and doc-check
* `compile` - group - run compile-fail and bench-check and example-check
* not providing a parameter will run everything
Ci is using those when possible:
* `build` jobs now don't run doc tests and don't build examples. it makes this job faster, but doc tests and examples are not built for each architecture target
* `ci` job doesn't run the `compile-fail` part but only format and clippy, taking less time
* `check-benches` becomes `check-compiles` and runs the `compile` tasks. It takes longer. I also fixed how it was using cache
* `check-doc` job is now independent and also run the doc tests, so it takes longer. I commented out the deadlinks check as it takes 2.5 minutes (to install) and doesn't work
