Anders Evensen

Enabling and disabling features

4

Second attempt at probing with features. First attempt was #33. This PR adds two methods to `AutoCfg`: `enable_feature()`, which enables a nightly feature to be used in probes, and `disable_feature()`,...

Add `assert_err_eq!` macro

1

Introduces an `assert_err_eq!` macro, which asserts that a `Result` value is an `Err` and that its contained value is equal to the provided value. Also defines a `debug_assert_err_eq!` macro. This...

Testing `DeserializeSeed` using `serde_test`

1

Is there any way to test types that implement `DeserializeSeed`, as opposed to `Deserialize`, using the `serde_test` crate? I am looking for some equivalent of `assert_de_tokens`, but that would accept...

Adding [`brood`](https://crates.io/crates/brood) to the benchmark suite. I included the benchmark results as well. From the other PRs I looked at, it seems that's the common practice, although my machine will...

Larger schedules cause compile times to balloon

2

It seems that larger schedules cause large blow-ups in compile time. I just compiled a schedule with 3 components and 14 systems, and the whole thing took over 11 minutes...

Currently, adding or removing components requires moving the components from one archetype to another, which requires an allocation here: https://github.com/Anders429/brood/blob/master/src/archetype/mod.rs#L461 Since the components are canonically ordered, this can be optimized...

Using heterogeneous list tricks, we should be able to remove the dynamic type checking in these two spots: - https://github.com/Anders429/brood/blob/master/src/registry/sealed/storage.rs#L644 - https://github.com/Anders429/brood/blob/master/src/registry/sealed/storage.rs#L754 Instead, these can be checked at the type...

Viewing one of two views with `Either`

1

Just an idea I had this afternoon: There may be times where we want to have a view on exactly one of two component types. For example, a system could...

`World::try_reserve()` method.

1

Like `World::reserve()`, but returns an error instead of panicking when the additional allocation is unable to occur. This is currently blocked on the fact that [`alloc::collections::TryReserveError`](https://doc.rust-lang.org/std/collections/struct.TryReserveError.html)'s internals are unstable.

This could give way to running sub-queries or sub-systems, among other things. This would provide a subset of `World`'s methods, similar to how the query `Entry` access provides a subset...