weird.. not sure why CI can'd find the package to install. Aren't we rebuilding the env every run?

Ah, looks like it's not actually on conda-forge yet. May push up a recipe on of these days.

I opened an issue to get a conda package published some time ago, no reaction yet: microsoft/playwright-pytest#113

It isn't going to be that straightforward as the pytest plugin depends on Playwright which itself isn't release on conda-forge but on the Microsoft channel. I assume first Playwright needs to have a recipe on conda-forge?

Thanks for opening the issue @maximlt. Do you know if that's the only way of getting packages merged to their channel?

No :/
I guess for now I'd just change the environment.yml so that pytest-playwright is installed with pip.

Seems like an acceptable work around for us to merge this, what's your opinion @philippjfr?

Agree with that, it's also what we do in the Panel test suite.

Hmm, the clock example either seems flakey or broken, it's failing for me on #522 as well.

Ah okay, but the problem with pytest-playwright is not replicable anymore right?
I'll rollback my last commit, cause I thought it was weird the previous one didn't work 😅

Yeah, I think you can roll that one back.

Since Fabio approved this, I'll merge :) once the CI says it's okay again (just out of superstition 🔮 👀 )

So confused about pre-commit, locally I see no issues.

Okay, I think what happened here is that some tests actually run quite quickly and the playwright call that checks for the loading screen runs after it has already been executed.

@philippjfr I don't think this is a CI only problem, because an user is seeing this locally too. #528

Yeah, the tests as written are just quite flakey. Trying to improve that but it's basically a race condition.

I can reproduce this every time locally, independently of the tests. I'm just running the example.
It happens with alpha and with the local build, on FF and chrome.

I can reproduce this every time locally, independently of the tests. I'm just running the example.

To be clear what is "this" you are referring to? I think we're talking about different things.

The type annotations are indeed still an issue but we don't have any tests checking them anymore.

Ah, sorry, @philippjfr.
I'm talking about this issue: #528. I thought it was the culprit for the CI crash.
What are you talking about?

I understand that you're making the tests sturdier, but I thought we were also fixing this issue 😅 sorry.

Trying to fix this weird test failure:

https://github.com/pyscript/pyscript/runs/6989466030?check_suite_focus=true#step:7:195

So we are talking about the same thing :)
I'm also trying to understand this.

I'm now hoping it was down to the custom (and pointless) <py-config> that was in the simple_clock.html.

Tests passing 🥳 Maybe let's restart the test one more time to make sure it wasn't just a fluke.

Hum... I'm not sure, honestly.
Strangely enough, the example wasn't working for me, but now it is, just like now it is for the CI. (EDIT: And I was running the local build version)
But I still have the <py-config> tag...
I'm not sure if we fixed it, but I can't reproduce it anymore.

Weird, always worked for me locally. Just very confused by the whole thing.

Well, I'll re-run the tests and if it's good we merge this.
I don't think it's related to the contents of this PR anyway but rather an issue with the example.
Thank you for your effort here Philipp! 🥳