324396 - Middle-click that crosses focus outline does not close tab

Status: RESOLVED DUPLICATE of bug 241622

(Core :: XBL, defect)

Description

[Jesse Ruderman]

Steps to reproduce:
1. Push the middle mouse button down on the active tab.
2. Mouse the cursor so it crosses the dotted focus outline (but does not leave the tab).
3. Release the middle mouse button.

Result: Tab does not close.

Expected: Tab closes.

This is a recent regression, probably from bug 308396.

Comment 1

[Martijn Wargers (dead)]

Is this still an issue now that bug 317375 has been fixed. Bug 317375 has made outlines transparent to events, so I guess this could be fixed by that.

Comment 2

[Martijn Wargers (dead)]

Sorry, I misunderstood the bug in comment 1.
This regressed between 2006-01-18 and 2006-01-21, so indeed likely a regression from bug 308396.

Comment 3

[Martijn Wargers (dead)]

Attachment: testcase

So basically with the fix for bug 308396, I think we now have this situation.

Comment 4

[Martijn Wargers (dead)]

So I guess this is a xbl bug.
However, bug 326851 is related, if bug 326851 is a genuine bug, then a fix for that bug would also fix this bug, I think.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Attachment: Same as just HTML, no XBL

So as HTML, I would NOT expect a click here (since the mousedown and mouseup are happening on two different nodes).

Should XBL really behave differently?  If so, then the ESM will need to deal, since that's where we generate our click events from mousedown and mouseup...

But I'm not convinced that XBL should behave differently in this case.  Also, I'd suggest filing a front-end bug on this stuff too, since there's no way I want to be touching the ESM on the 1.8 branch, even if we do decide to make the behavior different in the XBL case.  If the front-end folks care about the problem, they may want to consider a front-end-only change to fix it.

Comment 6

[Jesse Ruderman]

Why should whether a node sees a click event depend on whether the mouse moved between two of its children during the click?

Comment 7

[Martijn Wargers (dead)]

Boris, I was thinking in the line of bug 53007, except in this case I thought something like: "Need to suppress (retarget) mousedown/mouseup of anonymous content".

Your html testcase is what I filed bug 326851 about. Note that IE6 does fire a click event in this case (and personally, I think it makes sense).

Comment 8

[Boris Zbarsky [:bzbarsky]]

> Why should whether a node sees a click event depend on whether the mouse moved
> between two of its children during the click?

How do you define a "click"?  That's all there is to it.

Retargeting of the DOM mouseup/mousedown events won't help.  They're not used for click event generation.  Again, the magic's in the ESM.

Comment 9

[Jesse Ruderman]

The ESM bug is the same as bug 241622.  We probably don't need a frontend fix now that tabs don't have focus outlines all the time.