Leif Åstrand

> So, the question is: is it wrong to have LEADING listener executed twice (and we should fix this in this PR), or are the test assumptions that does not...

> Do you mean the tests `src/test-gwt/java` folder? I have a faint memory of such assumptions in the `test-gwt` folder. I don't remember if it was also somewhere else.

As a simpler workaround, we could maybe just create a built-in JS function for that particular case so that the whole expression wouldn't have to be sent back and forth....

> Is it possible to distinguish between closing a tab or refreshing it via the Beacon handler? No. The trigger for sending a Beacon request is the `unload` event that...

Note that the `ConfirmPasswordValidator` in itself is not enough for the use case of a field that is validated based on the value of some other field. In addition, you...

I'm not sure about that workaround since one might still want a way to explicitly test the offline state of an application even when it's running on localhost. I assume...

One thought was to always assume being online for localhost if `@PWA` is not used, while looking at the `navigator.onLine` flag if `@PWA` is present.

Does anyone have any idea on what `HTMLBodyElement.eval` might refer to? Seems weird that something like that would be related to `sendBeacon`.

I suspect it would be difficult for users to understand the difference between `com.vaadin:vaadin` and `com.vaadin:vaadin-platform` dependencies. The "correct" name for the the `vaadin` dependency with the contents it had...

Semantically meaningful names matter in lots of situations. Otherwise, we could just name dependencies `com.vaadin:a`, `com.vaadin:b` and so on and in that way avoid spending lots of time discussing names....