daxpedda

My 2 cents: > * Should we cooperate with the [image-rs](https://github.com/image-rs) organization? Or should we let image-rs to use J40 as a dependency? We should let `image` use the crate...

> Circular dependencies are allowed. Maybe we weren't talking about the same thing, I just made a quick test: ``` error: cyclic package dependency: package `test-circular-1 v0.1.0 (/test-circular-1)` depends on...

> Or are you looking into making this unnecessary? Yeah, that was the original thought.

> Hmm, what's the point of making it unnecessary? "every frame starts with a `resize` call" is a good rule of thumb as-is, I can't really see a reason to...

> is passing the size for each frame somehow a problem or inconvenience? It isn't. It's why I said this is **a very good** alternative. But it doesn't have a...

> I'm not trying to be adversarial here. All good! > I just don't know why you would want to elide the `resize` call in the first place. I'm not...

I didn't really dispute anything you are saying here. But as I pointed out in https://github.com/rust-windowing/softbuffer/issues/106#issuecomment-1693339385, eliding the `resize()` call is only an indirect goal, my point is to prevent...

Man, this issue is getting really off-topic. I would also prefer using a shared 2D vector type, but I don't like using an array instead. Indeed I would be more...

> Transparency support is somewhat special. We can't really convert RGBA to RGBX and have that to work as expected (but can do the reverse). The application needs to know...

> The complication there is format conversion. I'm in favor of doing no conversion at all. We take what the user supplies and that's it. The user has to supply...