fix: break if no result for getFormState

[zfben]

Changes

Fix when no this.result.

Testing

It doesn't change api, just check the property.

Docs

Not affect users' behavior, so don't need to update docs.

[changeset-bot]

🦋 Changeset detected

Latest commit: 45f3753

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[ematipico]

Can we add a test that shows we fixed a bug?

[zfben]

Can we add a test that shows we fixed a bug?

I don't know how to add a test case to reproduce this bug, it breaks my project and I found the code causing it.

[florian-lefebvre]

@zfben can you create a minimal reproduction for this? This way we could extract a test case from it

[zfben]

@zfben can you create a minimal reproduction for this? This way we could extract a test case from it

I forgot how it happens, just remember that render a page with just a react component and in middleware mode, it breaks. but it works when using client:only.

My project has skippied this bug, so i don't know how to reproduction it again.

[florian-lefebvre]

Well I'm going to close this PR then. Thanks for trying! If you ever face this problem again, please open an issue

[zfben]

@florian-lefebvre Can you reopen this PR? The bug happens again.
But I still can't create a minimal reproduction and my project is too complex.

[ematipico]

@zfben, if we can't reproduce it and create a test for that, we don't know what we're fixing.

You also didn't do anything for testing it. Unfortunately we can't accept this PR

[zfben]

@zfben, if we can't reproduce it and create a test for that, we don't know what we're fixing.

You also didn't do anything for testing it. Unfortunately we can't accept this PR

@ematipico
First, the original code was not written rigorously - it directly destructured an object without validation. I made this statement more rigorous. This isn't a case of making no corrections.

Second, I spent several hours trying to create minimal reproducible code but ultimately failed. And I cannot make my private project public for you to investigate the cause.

So for an issue that's hard to reproduce due to non-rigorous code, isn't it better to modify it into a more rigorous implementation?

[ematipico]

As maintainers, we are responsible for testing the code (even manually is fine), explaining to our users if they are affected by the bug and telling them if they should update and avail of this fix.

We can't hold ourselves accountable if we don't understand the bug or even test it against a project affected by it.

We aren't minimizing your efforts in fixing your bug, we're asking you to understand our position.

[zfben]

As maintainers, we are responsible for testing the code (even manually is fine), explaining to our users if they are affected by the bug and telling them if they should update and avail of this fix.

We can't hold ourselves accountable if we don't understand the bug or even test it against a project affected by it.

We aren't minimizing your efforts in fixing your bug, we're asking you to understand our position.

@ematipico Would this work? I can record a video demonstrating the bug occurring in my project, and then record another video showing that this fix indeed resolves the bug? It's not that I don't want to use test cases to more clearly demonstrate the cause of the error - rather, this bug is extremely difficult to reproduce in isolation, yet it has occurred multiple times in my project.

[zfben]

Additionally, this bug typically occurs after compilation, so some people might not have noticed it. And while the bug can be bypassed using client:only (which is what I'm currently doing, though it leads to decreased performance), it's possible that others who encountered the same issue have also used this workaround.