Document the supported browsers and general strategy

[romainmenke]

proposal:

• support 10 years of browser versions
• track Blink, WebKit, Gecko, Trident, EdgeHTML
• do one semver major each year
  • drop all polyfills that are't needed anymore
  • update tested browsers
  • ideally in the fall after Apple ships its latest OS's

---

Today the supported browser versions would be:

• IE 11 (because Edge 12 shipped in 2015)
• Chrome 38
• Firefox 33
• Safari 8

This implies:

• no more android specifics
• dropping IE 9 and 10
• dropping Opera Presto (if we still supported that)

Next year we could drop IE completely.

[romainmenke]

@fd, @mhassan1,

Currently CI tests are very flaky.
I haven't spend enough time investigating to say anything with absolute certainty, so to following is part conjecture.

There are two leading causes:

• the current browser set is aging and Browserstack isn't supporting such old versions very well
• more and more polyfills have been added over time, some very large

We've reached a tipping point where the current test setup and the surrounding infra just can't handle the amount of client side JS we try to run and test.

I think the setup could run fine again by removing some of the larger and older polyfills. I also think the setup could run fine again if we remove support for a large chunk of older browser versions.

I would really like to be able to take some time and fix what exists today, but I don't currently have the capacity for such an effort. Things are very busy at work and I don't think I can do much meaningful work in short sprints in between other work.

---

The timing might be right to take a relatively large jump forwards and drop support for a bunch of older things?

I kinda like the idea of:

• not having to support IE
• gaining access to JS features like classes, const, ...
• ...

I think this would also make it easier to add newer polyfills which often use more modern JS language features.

This could also make the project more accessible to a new generation of persons interested in polyfills.

Thoughts?

[mhassan1]

CI flakiness doesn't bother me too much. I don't think it is a strong reason for us to drop support for older browsers. If we need to, we can add more robustness to the pipeline (e.g. via more retries), but re-running flaky tests isn't so burdensome, at the moment, since we aren't making so many changes all the time.

It also doesn't seem like flakiness is necessarily related to old browsers; for example, here is a failure in modern Chrome: https://github.com/mrhenry/polyfill-library/actions/runs/12128696356/job/33818351286.

I'm okay with us dropping support for older browsers for other reasons, if we want.

[mhassan1]

Here's a PR that should improve CI flakiness: #89.

[romainmenke]

CI tests seem much more solid now, thank you @mhassan1 for working on this!
I am going to close a couple of these comments as resolved to bring this issue back to it's intended topic.