We should announce this and note it in the release notes.

Yes. Let's remember that we still insist on providing an adequate user experience for non-JS browsers, so dropping support for IE8 does not mean users on IE8 would be left in the dark.

I'd like to clarify that, given current traffic of IE8, I think it would be a mistake to tolerate runtime failure in IE8.

Reader experience for IE8 users should at least remain "basic" as defined at https://www.mediawiki.org/wiki/Compatibility.

I agree we should drop support for Grade A experience in IE8. Or more precisely, require ES5. Which I've been advocating for a few months now as an objective to achieve by the end of this calendar year 2015.

This means that server-side interaction (e.g. HTTP level), and content and layout accessibility (HTML/CSS) remains supported.

Change 252383 had a related patch set uploaded (by Jforrester):
[DONOTMERGE] startup.js Move IE8 from Grade A to Grade C

https://gerrit.wikimedia.org/r/252383

When would you expect this to happen?

In T118303#1798614, @Johan wrote:

When would you expect this to happen?

January.

https://lists.wikimedia.org/pipermail/wikitech-l/2015-November/083964.html

Wouldn't this implicate at the same time, that we can't fully support HTML5 Elements without extra <div> wrappers for a long(/eternal) time(s)?
Because in IE 6-8 HTML5 elements can hold children and are affected by CSS only by help of JavaScript.

Or would it be a special treatment for a special friend out of Grade C browsers to include one JS in a conditional comment in head exclusively for IE with JS enabled?

// Append JS shiv for supporting HTML5 elements in IE<9 @media screen 
$min = $this->getRequest()->getFuzzyBool( 'debug' ) ? '' : '.min';
$out->addHeadItem( 'html5shiv',
	'<!--[if lt IE 9]><script src="'
		htmlspecialchars(
			$this->getConfig()->get( 'LocalStylePath' ) .
			"/{$this->stylename}/resources/vendor/html5shiv{$min}.js"
		) .
	'"></script><![endif]-->'
);

In T118303#1855079, @Volker_E wrote:

Wouldn't this implicate at the same time, that we can't fully support HTML5 Elements without extra <div> wrappers for a long(/eternal) time(s)?
Because in IE 6-8 HTML5 elements can hold children and are affected by CSS only by help of JavaScript.

Or would it be a special treatment for a special friend out of Grade C browsers to include one JS in a conditional comment in head exclusively for IE with JS enabled?

Given that we already have this problem with IE6 and IE7 users (for over a year now), I don't think this issue should block the problem. I created T122965 to discuss this.

Change 252383 merged by jenkins-bot:
startup.js: Move IE8 down from Grade A to Grade C

https://gerrit.wikimedia.org/r/252383