Remove 'static requirement on run

[kchibisov]

There's no need to force the static on the users, given that internally some backends were not using static in the first place.

• Tested on all platforms changed
• [-] Added an entry to CHANGELOG.md if knowledge of this change could be valuable to users
• Updated documentation to reflect any user-facing changes, including notes of platform-specific behavior
• Created or updated an example program if it would help users understand this functionality
• Updated feature matrix, if new features were added or implemented

[notgull]

This would work for async-winit. In fact, it would work very well, since we could remove the 'static bound on the future in block_on.

However, it looks like this might be unsound on web to my uninformed eye,

[kchibisov]

@daxpedda @madsmtm @rib

I guess we can't do that for a reason that some backend no-return basically. However, maybe we should just give up and say that users should use spawn/run on web/ios respectively?

[madsmtm]

I think the macOS backend is currently unsound if 'static is not set, since we may not be clearing the callback consistently.

Though that should be fixable.

Re if the backend doesn't ever return, then not having 'static is still fine.

[kchibisov]

@madsmtm the thing is that we return, but the API consumes the EventLoop so the user can't rerun things, so EventLoop itself is dropped and users can't create it again.

@daxpedda I think I've missed something? Because it seems like other code is bound to 'static?

@madsmtm I was mostly curios about ios.

[daxpedda]

This should work fine for Web.
Rust actually just added support for the exception handling proposal, but only with no_std, which wasm-bindgen doesn't support. But when it actually hits run() on Web will be even weirder then before and actually breaks without 'static. I can add a check to make run() not available if exception handling is enabled.

@daxpedda I think I've missed something? Because it seems like other code is bound to 'static?

I can fix that.

[madsmtm]

I'm not sure I understand you, but I was trying to say that the change should be sound on iOS (even though it doesn't compile as-is).

[daxpedda]

@daxpedda I think I've missed something? Because it seems like other code is bound to 'static?

I can fix that.

Required an unsafe conversion, but works.

[kchibisov]

Required an unsafe conversion, but works.

Yeah, that's what I was afraid of, so I left it as is, unless a backend maintainer approves.

[kchibisov]

@madsmtm you've replied about macOS initially though, not ios though. I guess your reply about ios was indirect though.

Does it matter for macOS how you cleanup, if you basically drop the EventLoop, like 'static shouldn't matter once you literally call drop on the entire loop in the end of run invocation?

[daxpedda]

For Web this works, but probably not useful and I would argue makes run() more and more problematic.

We can have another discussion about whether we should leave run() on Web or not in a different place and time.

[kchibisov]

I guess I'll do the beta 2 with it like that, and then we could probably gather some feedback whether we'd really need it to keep that way.