1079131 - [highlighter] Highlighter doesn't stop tracking the mouse and JS error in e10s

Status: RESOLVED WORKSFORME

(DevTools :: Inspector, defect)

Description

[Patrick Brosset <:pbro>]

STR:

- open an e10s window
- open the inspector
- click on the element picker icon
- hover elements on the page
- then, with the element picker mode still ON, move your mouse over the inspector panel (this should highlight elements in the page)
- click on one of the nodes in the inspector (the element picker should still be ON)
- then move your mouse back over the page and try to select an element by clicking it

=> Expected: the element should be selected in the inspector and the element picker mode should be OFF

=> Actual: the picker mode remains ON

Then clicking on the element picker button to try and turn it off throws an error in the console:

console.error: 
  Message: TypeError: target.removeEventListener is not a function
  Stack:
    exports.HighlighterActor<._stopPickerListeners@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/highlighter.js:237:5
exports.HighlighterActor<.cancelPick<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/highlighter.js:255:7
actorProto/</handler@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/protocol.js:955:19
DSC_onPacket@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/main.js:1407:15
ChildDebuggerTransport.prototype.receiveMessage@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/transport/transport.js:708:5

exports.HighlighterActor<._stopPickerListeners@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/highlighter.js:237:5
exports.HighlighterActor<.cancelPick<@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/actors/highlighter.js:255:7
actorProto/</handler@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/protocol.js:955:19
DSC_onPacket@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/server/main.js:1407:15
ChildDebuggerTransport.prototype.receiveMessage@resource://gre/modules/commonjs/toolkit/loader.js -> resource://gre/modules/devtools/transport/transport.js:708:5
console.error: 
  Protocol error (unknownError): TypeError: target.removeEventListener is not a function

Remy Sharp of jsbin has been testing the devtools recently and stumbled on this issue while recording this video: https://www.youtube.com/watch?v=4muwlibj6zA

Comment 1

[Patrick Brosset <:pbro>]

Of course, to make things fun, this doesn't seem to always happen. If it doesn't, keep playing with the highlighter, inspector and pick mode. It's usually easy enough to reproduce the problem.

Comment 2

[Patrick Brosset <:pbro>]

And ... now I can't reproduce anymore.

Comment 3

[Patrick Brosset <:pbro>]

Remy, if you have better steps to reproduce that lead to the problem 100% of the times, that'd be great.
The randomness of it makes me thing this is a timing issue.

Comment 4

[Heather Arthur [:harth]]

I can reproduce this sometimes. I added some logging and the problem is when we get the docShell's chromeEventHandler it doesn't implement nsIDOMEventTarget:

Good:
----
docShell.chromeEventHandler [xpconnect wrapped nsIDOMEventTarget]

Bad:
----
docShell.chromeEventHandler [xpconnect wrapped nsISupports]


I personally don't know how it can get in that state, but bug 992778 was mentioned in a related comment, so I'm CCing Alex who fixed that.

Comment 5

[Alexandre Poirot [:ochameau]]

I'm not able to reproduce it easily. (Not being able to reproduce it after 10 times on various websites)

I would say that docShell.chromeEventHandler exposing only nsISupports may tell us that this docshell is destroyed.
That would be worth trying to check for docShell.contentViewer / docShell.contentViewer.DOMDocument / docShell.contentViewer.DOMDocument.readyState.

Comment 6

[Mike Ratcliffe [:miker] [:mratcliffe] [:mikeratcliffe]]

I can't reproduce this at all with the patch for bug 985597 applied (highlighter in canvas frame).

Comment 7

[Chris Peterson [:cpeterson]]

These DevTools bugs should probably block e10s from riding to Aurora.

Comment 8

[Chris Peterson [:cpeterson]]

devtools bugs will block the e10s merge to Aurora, but blassey would like them to be tracked by dte10s meta bug 875871, not the tracking-e10s=m6+ flag.

Comment 9

[Mike Ratcliffe [:miker] [:mratcliffe] [:mikeratcliffe]]

Patrick: I still can't reproduce this... are we okay to close it?

Comment 10

[Patrick Brosset <:pbro>]

Yep, can't reproduce either. Let's close it.