[SpeedDialIcon] Fix test for react 16.12

[eps1lon]

Started failing on the react-next pipeline (https://app.circleci.com/jobs/github/mui-org/material-ui/125891)

I'll remove the resolutions before merging. Just making sure the react-next build is green again. It is green with react@next: https://circleci.com/workflow-run/ecebfcea-4f90-417c-a871-815808526b5f

[mui-pr-bot]

No bundle size changes comparing a54729e...9533904

Generated by 🚫 dangerJS against 9533904

[oliviertassinari]

Do you know why it fail? Could it be a regression on React side (that might be fixed by the time a new version is released)?

[eps1lon]

Since enzyme simulates react internals any internal change of react internals might cause a regression in enzyme. We have a history of having issues with React.memo and enzyme#find. It's one of the reasons we're switching from enzyme.

[oliviertassinari]

Fair enough, so in the long term, it doesn't worth looking at what changed, it's better to use testing library 👍 .

[eps1lon]

Fair enough, so in the long term, it doesn't worth looking at what changed, it's better to use testing library .

Yeah. There was some change related to react-is in 16.12 and I do remember brian saying it was ok with enzyme. There was also a related devtools change so I'm pretty confident react had no public regression. Enzyme is a different story though.