Something got broken in 1.8.1 or 1.8.0

[mindplay-dk]

Here's a little experiment I've been working on:

https://codesandbox.io/s/fre-demo-iefch

I wanted to see if it was possible to implement React's class-based Component API using hooks.

This was working fine until release 1.8.1 - so the last release that worked for me was 1.8.0.

Try changing the dependency version from 1.8.0 to 1.8.1 in the sandbox, and all the calls to setState for some reason stops working. 😶

I suspected this had something to do with my recent change inlining the setter-function - but I've tried reverting that change, and it still doesn't work, so that's not why.

Here's a diff between 1.8.0 and 1.8.1.

Can you spot the problem in these changes?

Whatever it is, it's obviously not covered by tests yet...

[yisar]

How about 1.8.3

[mindplay-dk]

How about 1.8.3

All versions after 1.8.1 affect that example.

Here's another simpler example:

https://codesandbox.io/s/fre-demo-uxuic

This one starts to break with version 1.8.0 already, so the bug may have been partially introduced there.

Watch:

All of a sudden, it's like the scheduler stops accepting new tasks or something?

If you look in the console, you can see event handlers are still firing - it's the calls to the useState setters that aren't triggering updates.

Could be related to this maybe?

[yisar]

This looks like 1.8.0 is wrong, in version 1.8.3 I removed [] and it worked

const component = useMemo(() => new Component(props));

[mindplay-dk]

Okay, we still have missing tests for useEffect with different deps arguments - all the tests use an array right now. If this is the problem, I guess that's why the tests didn't catch it.

That might explain the problem with the first example.

But according to useMemo docs:

If no array is provided, a new value will be computed on every render.

That's not what I want in the first example - I'm passing [] because the function has no dependencies and should be executed only on the first render.

And the second example? The error is periodic, as you can see in the GIF - so probably has something to do with the scheduler? Might be two separate problems.

Gotta get the rest of the tests in place...

[yisar]

I'm sorry, your code is correct, I just debugged it, this is indeed my fault, I will fix it as soon as possible, thanks!

[mindplay-dk]

Okay, cool - I'll try to get the missing coverage for hooks today, so we can make sure this stuff doesn't break again 🙂

[yisar]

It seems that I have fixed this bug, and then it should be #80

[mindplay-dk]

I'm sorry, I think you misunderstood.

The test was correct - it failed because the effects weren't being triggered correctly.

Now the test is wrong: the same cleanup and effects should not execute more than once!

[yisar]

Can you give me an example with sandbox? the test I am not sure weather it is correct.

[yisar]

In fact, the bug here has nothing to do with useeffect, but all my debugging results are the same as react. I need an example to show the difference between react and fre, not the document or the test ,Thank you very much

[yisar]

Let me explain again. I'm not sure whether the test case is correct or not, but I won't say it's wrong. I have to get an example that can be debugged, or I will maintain the status quo. If the example proves my mistake, it's not difficult to fix it, please forgive me.

[mindplay-dk]

I have to get an example that can be debugged, or I will maintain the status quo.

That's what the tests are for 😉

I guess from the HTML files in the demo folder that you're debugging from an HTML page in a browser?

When you run the tests under node, you can connect (for example from VS Code or WebStorm, whatever you're using) to the debugger, set breakpoints, inspect variables, run individual tests with a click, and so on - it's definitely worth learning. 🙂

[mindplay-dk]

With your last commit, I can confirm that my sandbox example is now working correctly!

It took me a while to figure out why the test is passing with your change - it looks wrong, but now I see the error in my test! PR on the way! 😁

[mindplay-dk]

For the record, the weird problem I showed above where the scheduler seemed to die - this also no longer seems to occur, so these were probably related. 👍

[yisar]

I published v1.8.5👏