[CLEANUP beta] Update browser support per RFC #685

[kategengler]

https://rfcs.emberjs.com/id/0685-new-browser-support-policy

May be worth considering an update to the policy to drop Safari versions with < 0.25% share.

[chriskrycho]

👍🏼 This looks good. I'd likewise be in favor of dropping older Safari with < 0.25% usage (I personally would be fine with a 1% cutoff but different folks may have different tolerance levels).

[chriskrycho]

To expand on that: >0.25% and not dead would still have us supporting mobile Safari 12, though, at 0.38%. This is almost certainly due to a "cliff" in device support (Apple's overall OS release support is long, tends to have those kinds of cliffs); mobile Safari 12.2–12.5 actually has higher usage than 14.0–14.4 (though lower than 14.5–14.8), so changing the policy to that wouldn't change what we target.

[kategengler]

The RFC is hardcoded at Safari 12, so if we do eventually want to drop, we need to consider/pay attention to it.

[chriskrycho]

Yep, I think if we revisit we should do it in terms of a % point, while preserving the notion of non-evergreen browsers not dropping support. The 12.x iOS Safari versions are now less than a third of the usage they were when we adopted the current policy, so I think it would be perfectly reasonable to bump the target—notably, even just targeting 0.5% would bump all the way up to 15.6. (I don't have a strong opinion that we should do that, but I do think we could do that!)

[wycats]

I agree that we should consider a usage-based cutoff for Safari (the one "non-evergreen" exception in our current RFC).

Might we also want to consider a cutoff based on official Apple support?

Catalina came out in 2019 and is no longer officially maintained, but as far as I can tell, we're still supporting the version of Safari that came with Mojave, which hasn't received an update since 2019.

iOS 12 (which came with Safari 12), on the other hand, is still limping along with security updates, but has been out of active support since 2019.

All of these details conspire to keep Safari 12 above 0.25% but consistently below 0.5%.

notably, even just targeting 0.5% would bump all the way up to 15.6.

In the case of Safari, would we want to say that the number is the cumulative number of users using <= the target version?

While each of 12 and 14, on their own, aren't above 0.5%, more than 0.5% of mobile users use a version of Safari "no newer than Safari 14." In the special case of non-evergreen browsers, isn't that the intuitively reasonable metric?