880022 - Ensure CustomizableUI.jsm doesn't break when there are no browser windows open

Status: NEW

(Firefox :: Toolbars and Customization, defect)

Description

[Blair McBride [:Unfocused] (UNAVAILABLE)]

Quoting from bug 873066 comment 14:
> However, we'd still
> need the code to handle the case where there are no windows - mostly due to
> OSX when there are no browser windows open, although it'd be possible to hit
> it in other ways too (even earlier in initialization, or when there's an
> add-on window open but no browser window).


Need to check if anything in Customizable.jsm breaks if there are no browser windows open when any API is called.

Comment 1

[Mike Conley (:mconley) (:⚙️)]

Removing the items from M7 that do not block us from landing on m-c.

Comment 2

[Justin Dolske [:Dolske]]

Not sure how to prioritize this. P5 without any known impact, but sounds like someone needs to look and see if that's actually the case. Are there any common scenarios where this might happen?

Comment 3

[Blair McBride [:Unfocused] (UNAVAILABLE)]

(In reply to Justin Dolske [:Dolske] from comment #2)
> Are there
> any common scenarios where this might happen?

Easy to get into this situation on OSX - close all windows, without closing the app.

Comment 4

[:Gijs (he/him)]

(In reply to Blair McBride [:Unfocused] from comment #3)
> (In reply to Justin Dolske [:Dolske] from comment #2)
> > Are there
> > any common scenarios where this might happen?
> 
> Easy to get into this situation on OSX - close all windows, without closing
> the app.

I've been testing this from the browser console, and it seems to work well. Even, say, removing a widget that has removable="false" doesn't break stuff, because the widget ID is removed from placements at that point, but re-added once a window opens (because buildArea realizes it can't actually be removed).

Should we add automated tests for this (OS X only, most likely?), or should we just mark WFM and carry on with life?

Comment 5

[Blair McBride [:Unfocused] (UNAVAILABLE)]

Ah, good to know :)

I think we should have tests on OSX for this, since it seems like something that would be easy to accidentally break.

Comment 6

[:Gijs (he/him)]

(In reply to Blair McBride [:Unfocused] from comment #5)
> Ah, good to know :)
> 
> I think we should have tests on OSX for this, since it seems like something
> that would be easy to accidentally break.

Unfortunately word in #developers was that having mochitests that deal with 0 open browser windows wasn't currently possible, so I've not looked into this further yet.

Comment 7

[Mike de Boer [:mikedeboer]]

So this means that this is impossible to do atm? P-?

Comment 8

[:Gijs (he/him)]

(In reply to Mike de Boer [:mikedeboer] from comment #7)
> So this means that this is impossible to do atm? P-?

No, it just means it's not trivial. We should still do this, and I'd even say it's somewhat important because we don't want to get things messed up if people close all their browser windows on OS X. It's just going to take work.