To differentiate between WebDriver classic (HTTP) and WebDriver BiDi the classic spec got a HTTP flag added:
https://w3c.github.io/webdriver/#dfn-http-flag

This flag would be needed for Multiple Session support (bug 1862018).

It looks like that I missed to update the whiteboard for this bug given that this bug is listed as required for M11 and after implementing bug 1884650 I would agree to have it in this milestone.

Try push for the soon upcoming patches:
https://treeherder.mozilla.org/jobs?repo=try&revision=0abad9524ceaa2934e0f0cb163c5ccce57864225

Backed out for causing mochitets failures in browser_Troubleshoot.js

• Backout link
• Push with failures
• Failure Log
• Failure line: TEST-UNEXPECTED-FAIL | toolkit/modules/tests/browser/browser_Troubleshoot.js | Schema mismatch, Error: Validation error: Object is not of the expected type: object=["[GFX1-]: glxtest: ManageChildProcess failed\n","[GFX1-]: No GPUs detected via PCI\n"], schema={"type":"object","properties":{"key":{"required":true,"type":"string"},"args":{"required":false,"type":"object"}}} -

This is a failure when getting the GfxInfo. A change that I've made is the following:

-        "moz:headless",
-        Cc["@mozilla.org/gfx/info;1"].getService(Ci.nsIGfxInfo).isHeadless,
[..]
+ ChromeUtils.defineLazyGetter(lazy, "isHeadless", () => {
+    return Cc["@mozilla.org/gfx/info;1"].createInstance(Ci.nsIGfxInfo).isHeadless;
+. });

So I've accidentally replaced the call to getService() with getInstance(), which clearly should fail. Still interesting that this is still working fine on other platforms than Linux and even Linux opt. I'll fix and push again.

https://hg.mozilla.org/mozilla-central/rev/374f26d432ea