Back

1. https://github.com/bevyengine/bevy/pull/15156

Ah nice, the light visibility rechecking PR is merge conflict free! It's now failing CI though: we've added a stricter check against platform-specific trigonometry implementations. Bah. Well, I'll let the author know. Not a big deal if it misses the release.

2. https://github.com/bevyengine/bevy/pull/16088

Oh good, a consistency and quality pass on how we handle texture atlases in both UI and 2D cases. This is a big part of why I like the release candidate process so much: it really forces us to polish and reconsider the choices we've made in the past few months. Glad to see that this is reviewed by Cart too!

Very reasonable set of changes, and well-made. I'm so grateful for Ickshonpe: they really focus on careful, well-motivated, self-contained fixes. Merging <3

3. https://github.com/bevyengine/bevy/pull/16111

A meshlet PR! Good to see that this got reviewed: I know Jasmine's been frustrated by how slow the turnaround is on it. Selecting the correct level of detail based on the error accumulated in the normals in addition to the positions fundamentally seems reasonable.

I do wish the upgrade had been split out, but eh, reviewing commit-by-commit shows what's going on :) Merging!

4. https://github.com/bevyengine/bevy/pull/16151

An extension to our UI example to demonstrate some shiny new features. Outlines, border radius, flipping, image sizing...

For the most part, this example is useful as a quick visual regression test. As you can see from the screenshot, it doesn't demonstrate anything resembling a useful UI to users!

I'd like to move this to our new testbeds folder, but this change is definitely valuable even without that. Merging and opening a follow-up issue.

5. https://github.com/bevyengine/bevy/pull/16157

When blending camera outputs by rendering them both to the same texture (TIL this is possible!), the behavior is wrong if one of them is set to be skipped.

There's good discussion about follow-up work, but I agree with the author: it's better to land the targeted fix now. Very straightforward bug fix, merging :)

6. https://github.com/bevyengine/bevy/pull/16188

A straightforward "move code in the rendering engine around so we don't waste work" PR. The PR description is excellent, covering why this change helps both improve flexibility and improve rendering performance in scenes with multiple 3D cameras.

This is a really nice small cleanup: it makes me happy to see rendering get this sort of fix. Merging.

7. https://github.com/bevyengine/bevy/pull/16202

A tiny PR from a new contributor, fixing a malformed comment in a random shader file.

Clearly correct, merging. It's still bizarre to me that the shader ecosystem feels so fragmented and immature compared to ordinary programming lanugages. The WESL group has plans there though, and whenever I've talked with the wgpu maintainers they've been super sympathetic to this perspective. Maybe there's a brighter future for rendering engineers :)

8. https://github.com/bevyengine/bevy/pull/16211

A neat little convenience API for working with `GlobalTransform` from a new contributor. Being able to access only one of the TRS components at once is reasonable and we can save a bit of work.

Mentoring this one was fun: the author was really patient and humble as they figured out the process of exactly how to contribute, what comments to leave, how to design the API and what sort of tests to write.

I'm pleased with the final result. Merging.

9. https://github.com/bevyengine/bevy/pull/16220

A small assets PR, prompted by editor work to load and save assets. `AssetWriter`, which abstracts over the IO interface across platforms (most important on web) for Bevy, can now create folders.

Very straightforward, and handy enough. Not used in 99.9% of games, but Bevy's useful for tooling and weird CAD-adjacent programs too :) Merging.

It's really nice to see the dogfooding pay off even in this really early stage.

10. https://github.com/bevyengine/bevy/pull/16228

A trivial PR from this morning: some of our debug messages for hooks were trying to return information that wasn't provided! Very much looks like a refactoring issue to me, and that PR had a bit of a tortured life so that's not surprising.

Annoying that we can't pass on that info, but that's a bigger conversation. Regardless, this PR is 110% an improvement over the status quo. Uninformative error messages are better than broken *and* uninformative error messages.

@piturnah Yeah, me too! Let me check the clippy repo...