Bug with group focus

[aless]

Hi again everyone,
we have been following this closely, and given the response here, relative to the overall minor consistency win, we have decided to revert this change, and to allow Toggling Group Focus elements from a click again.

Several notes about why we went with this revert:

• We had misjudged how broken this was in the first place: while we always saw this behavior to be broken in our tests, it turns out that under some circumstances, it was indeed working properly - I believe this depends on the implementation, and on the browser, because this is a more complex race condition than just “Group Focus hides before workflows run, so Toggle is broken”. Because this type of interaction makes sense from a UX standpoint, and because some people experienced it to not be broken (including us in a few places in the our own Bubble website when we built them), there was a much higher use of this feature than we expected, and this change affected many more people than we would have been comfortable pushing a silent breaking change for.
• While we expected this to only apply to new action creations, this also showed up in the issue checker and blocked deploys for what was essentially a mere recommendation, therefore this was a much larger breaking change than the tiny deprecation we thought it would be.

At this point we still recommend against using this action, under these circumstances, since it will look glitchy at the moment in some browsers - Group Focus elements will blink for a fraction of a second, then open again, instead of closing, when you toggle them closed.

Instead, the Two-step Show/Hide action set-up with conditions, which we have seen as a workaround, while less compact, should work much more reliably.

We are planning to tackle this as bug down the line, but it is a bit trickier to solve, because it relies on poorly defined behavior of what “should” run first between workflows and real-time page interactions.

We are also considering adding another level of issues that aren’t blocking - essentially warnings - to recommend against a particular pattern, while not preventing from deploying apps. This would have made this and other updates much smoother overall, so once we do this, smaller UX consolidations will be much less annoying to all users.

I hope that this sheds some light onto why we made the initial change, and why we ended up backtracking on it.

Sorry for the disruption, and happy building,
Best,