Better define what goes in the compat spec

[karlcow]

What is the issue with the Compatibility Standard?

There has been a bit of pressure in some issues these last couple of months on things that should be added to the compat specification.

Currently the spec says:

This standard describes a collection of web platform features that web browsers need to support for compatibility with the de facto web.

That's a good and simple definition.

I wonder if we should add a couple of words to the README.md to frame a bit more the decision criteria for adding something or not.

To reformulate what is written above, we could add to the README.md something along:

---

The Compat Standards describe a set of technologies that

1. Browsers need to implement to be compatible with the Web as-is.
2. No Working Groups have an intent to make it into their specs at the moment.

This is not a place to document interoperability differences in between browsers, or what needs to be solved in between browsers so that the Web would work better. This kind of work has to be done and published usually in WICG or Interop Investigation Areas.

---

One section of the compat spec is unfortunately going against this criteria. The User Agent strings section, which document the way the browser exposes the UA strings. It doesn't make the Web more compatible Web. I wonder if retrospectively it was a mistake to put it here. Maybe we should revise this too, to avoid the ambiguity described above.

[miketaylr]

I wonder if retrospectively it was a mistake to put it here. Maybe we should revise this too, to avoid the ambiguity described above.

I think I disagree, which is probably not surprising! Maybe it's the way that things are defined, but HTML has a very similar concept: https://html.spec.whatwg.org/multipage/system-state.html#concept-navigator-compatibility-mode

One option is to redefine things in terms of navigator compatibility mode?