Change 482783 had a related patch set uploaded (by Jdrewniak; owner: Jdrewniak):
[mediawiki/extensions/MobileFrontend@master] Drawer.js DOM inserted onShowDrawer instead of on initialization

https://gerrit.wikimedia.org/r/482783

Change 484776 had a related patch set uploaded (by Jdrewniak; owner: Jdrewniak):
[mediawiki/skins/MinervaNeue@master] Prevent black background from appearing on menu after clicking reference

https://gerrit.wikimedia.org/r/484776

After looking through the code, I think the regression might have been caused by use moving modules around during the refactor, and causing some CSS files to be loaded in a different order.

Judging from these two rules, they have the same specificity and it looks like .primary-navigation-enabled' might have been responsible for overriding the .has-drawer rule with a different background color.

The patch above fixes that.

It does not solve the underlying issue that the Drawer appends a class to the HTML and never removes it, or that the Drawer DOM itself is never actually removed from the document after closing a drawer...

I think those underlaying issues are concerning but deserves their own ticket.

Change 484776 merged by jenkins-bot:
[mediawiki/skins/MinervaNeue@master] Prevent black background from appearing on menu after clicking reference

https://gerrit.wikimedia.org/r/484776

@Jdrewniak looking through this again, specifically in the context of T214049 I really think we should also be updating Drawer.prototype.onHideDrawer to also remove the class has-drawer it adds in postRender. Patch to follow.

Change 485972 had a related patch set uploaded (by Jdlrobson; owner: Jdlrobson):
[mediawiki/extensions/MobileFrontend@master] Make Drawer's contract clearer

https://gerrit.wikimedia.org/r/485972

I really think we should also be updating Drawer.prototype.onHideDrawer to also remove the class has-drawer it adds in postRender. Patch to follow.

I didn't want to touch the .has-drawer class because it's responsible for adding transitions to the overlay. But looking at your patch, it seems like we can still maintain those transitions, the only one I don't see is the opacity fade-out transition when the drawer closes.

I don't think the minHideDelay is used for animation. I think that's a hack to get around attaching the window.click event in the onShowDrawer method. In a previous patch (which I just slowly walked away from... ) I attempted to add another timer for the hide method. It's kind of brittle because I hard-coded the duration at 250ms, which is a value that's actually changed in CSS.

Anyway to your point about the rules in toast.less, I don't think those are clashing styles, I might be mistaken but I think those are the actual styles responsible for sliding the drawer in and out.

The styles responsible for fading the background in and out are actually in drawer.less:

// Darken the content to highlight the drawer.
.has-drawer {
	background-color: @colorGray1;

	> *:not( .drawer ) {
		opacity: 1;
		.transition( opacity .25s linear );
	}

	&.drawer-visible > *:not( .drawer ) {
		opacity: 0.6;
	}
}

So removing the .has-drawer class also removes the transition on > *:not( .drawer ) (i.e. everything on the page except .drawer).

I guess if we wanted to keep that transition we would have to create a class like

.animations > *:not( .drawer ) {
    .transition( opacity .25s linear );
}

But that still looks kind of weird because background color changes from grey to white.

Change 485972 abandoned by Jdlrobson:
Make Drawer's contract clearer

Reason:
To be done on a different patch

https://gerrit.wikimedia.org/r/485972

I chatted with Jan after standup, and we agree that it makes sense to tackle further changes in a new task which I will flesh out today.

Edward please procede with QA

This passed with a ChromeOS virtual machine. I'm going to check it with an emulator that allows me to test with the same resolution as a Pixel Slate.

Change 482783 abandoned by Jdlrobson:
Drawer.js DOM inserted onShowDrawer instead of on initialization

Reason:
Larger conversation continuing in https://gerrit.wikimedia.org/r/524314

https://gerrit.wikimedia.org/r/482783