Unfortunately this PR has caused some Google internal failures. I'm going to fix them (but probably not very soon since I have a lot on my hand). Also this change will probably count as breaking changes and I'll write a migration guide. Therefore don't expect this PR land very soon.

Unfortunately this PR has caused some Google internal failures. I'm going to fix them (but probably not very soon since I have a lot on my hand). Also this change will probably count as breaking changes and I'll write a migration guide. Therefore don't expect this PR land very soon.

Why will it count as breaking change? I don't see any changes in the API itself

@dkwingsmt this should not be a breaking change. yes, updating scroll physics will cause some golden files to change heavily, but fixing this is more important than keeping those goldens the same.

I think the idea is that developers do not expect their apps to behave differently after a version update. Silent changes like this are even worse than compile errors. A migration guide should be provided to let developers know how to calculate new values to achieve the previous effect.

No that is absolutely not our policy. We need to be able to update things like scroll physics

By our policy,

Additional test suites that we have been allowed to run but that are not public. (Notably, Google allows us to run several tens of thousands of proprietary tests on each commit.)

This PR breaks a few tests in Google tests. Shouldn't it count?

If you changed something unrelated and a bunch of goldens broke, it is a good sign you introduced a bug yes. But in this case, you are intentionally fixing the spring simulation formula - so its expected that golden tests which perform something like a fling will change. Arguably these tests should not depend on the exact physics, because at any rate we must be able to make updates to physics to match the platform.

We are absolutely allowed to update golden files. Unless it is a major change (i.e. light colors become dark) we do not need to flag it.