Changed EventLoop to give a result instead of panicking

[lemonlambda]

• Tested on all platforms changed (Tested control_flow and cursor_grab on linux)
• 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

I wanted to because if creating a EventLoop fails because not on main thread then you could match that and do EventLoop::new_any_thread() on linux you can at least.

[filnet]

You can test if you are on the main thread yourself and call the appropriate new function (when on linux).

[dhardy]

Seems sensible to me — I already have to handle errors from other things constructed alongside the EventLoop.

[dhardy]

Check doc on this method

[lemonlambda]

I am guessing you mean change the docs for this one, and the example

[lemonlambda]

You can test if you are on the main thread yourself and call the appropriate new function (when on linux).

To me this just feels like trying to get around the issue of panicking instead of just making it return a result instead of panicking. Which returning a result feels cleaner to me.

[madsmtm]

Let's merge #2137 first, that has a fair implication on this

[lemonlambda]

Let's merge #2137 first, that has a fair implication on this

#2137 Has been merged do you think now this has a chance?

[madsmtm]

Yeah, I think it makes sense to do this change (but please rebase first).

However, since it is probably quite rare that people will be needing this, maybe we should consider EventLoopBuilder::try_build/EventLoop::try_new?

[lemonlambda]

I have fixed the conflicts in the different branches, and it works fine from testing custom_events and control_flow

[madsmtm]

Could you put those changes in this branch (by using either git rebase or git reset --hard, and then git push --force-with-lease afterwards)

[lemonlambda]

Could you put those changes in this branch (by using either git rebase or git reset --hard, and then git push --force-with-lease afterwards)

I have done this now

[lemonlambda]

For some reason I was on the wrong branch, now I am on the right branch

[lemonlambda]

I have made those changes

[lemonlambda]

@madsmtm Any changed I need to do now or is it good?

[madsmtm]

Sorry, I don't have much time lately.

But CI doesn't pass, so that's at least missing.

[lemonlambda]

Sorry, I don't have much time lately.

But CI doesn't pass, so that's at least missing.

Cl?

[maroider]

CI?

Continuous Integration.

In our case, we make sure everything is formatted correctly and compiles on all platforms (there's also a test suite, but I believe that one is relatively small).

You can see the things that failed at the bottom of a PR's page.

We need all of these checks to pass in order to merge.

[lemonlambda]

Why are they failing?

[maroider]

You can click on the "details" button to find out

[lemonlambda]

You can click on the "details" button to find out

it looks like cthulu is speaking in my ears

[maroider]

Some of winit's public structs (like EventLoop and Window) are shims to a platform-specific struct. Thus, I can without looking too closely at the specific error messages be fairly confident that most of the errors are type mismatch errors stemming from a change in the public API without a corresponding change on every backend.

[lemonlambda]

Why isn't it telling me in my IDE there are these type errors

[maroider]

Because your IDE only runs checks for a single target triple (presumably i686-unknown-linux-gnu, since that's the only triple I see where CI passes), while we try to run checks for every supported target triple. If you have the appropriate targets installed through rustup, you can check for these errors locally with cargo check --target=<target-triple>. Your IDE should also have some way of configuring the target triple used when running checks.

[lemonlambda]

I do not feel like making the required changes as of right now to finish this pull request so I will now close this. (I don't know if other people can reopen pr's but if someone else wants to finish this they can)