Refactor back to Popper

[RobbieTheWagner]

There is a Popper beta in the works that is supposedly very small and performant. We should give it a try and see if it fixes some of the issues we are having with Tether.

[FezVrasta]

FYI yesterday we released Popper 2 (stable)

[chuckcarpenter]

@FezVrasta having a tough time getting this completely working w/ v2. My main issue is the window scrolls back to top and then to the target. It seems that the position:absolute is applied first and then the calculations for the transform position.

And then

I imagine I need to leverage a modifier here, but not clear where to start. If you have any ideas I'd appreciate it!

[FezVrasta]

Have you tried to set the position: fixed; top: 0; left: 0; as default style for the elements? This should prevent the page from scrolling automatically up.

Alternatively you may disable any focusable element inside the popper before popper positioned it

[chuckcarpenter]

@FezVrasta would I do that by changing applyStyles modifier? Seems like that's where the initialStyles values are that get written inline first. I tried creating a custom one beforeWrite that sets state.elements.popper.style, but doesn't seem to change the default value.

[chuckcarpenter]

A little more context, applying default styles on the element doesn't work adding in a custom modifier to set values isn't working either. If I change the applyStyles effects function for the initial values, then it works. Is that my only option @FezVrasta?

function effect(_ref2) {
  var state = _ref2.state;
  var initialStyles = {
    position: 'fixed',
    left: '0',
    top: '0',
    margin: '0'
  };

...

[chuckcarpenter]

This fixes it, but would be great to know if there's a less intrusive path. #801

[FezVrasta]

I see, probably we should do that in Popper, would you mind opening an issue on the repo?

[atomiks]

Does the scroll problem happen on Firefox or Chrome 79? It's come up twice recently (including Popper v1) which makes me think Chrome 80 did something?

[RobbieTheWagner]

@atomiks Chrome 80 broke array reduce I think. Not sure if that applies here https://www.reddit.com/r/webdev/comments/f26mbn/chrome_80_introduces_a_new_bug/

[atomiks]

We use reduce everywhere in Popper 😱 but yeah I don't think that would be related. It seems like it was reported back in mid 2019 so I guess not...

[RobbieTheWagner]

@atomiks the bug effects Chrome 80-82 I believe. We hit errors with reduce just the other day, definitely still an unresolved bug in Chrome.

[chuckcarpenter]

@atomiks I can confirm this is happening in FF 71.

[RobbieTheWagner]

We've got popper v2 fully working now I believe, so closing this. Thanks everyone!