698662 - error in console: this._stringBundle is undefined

Status: RESOLVED DUPLICATE of bug 636947

(Firefox :: Sync, defect, P2)

Description

[Dietrich Ayala (:dietrich)]

points to this line in browser.js:

onSyncFinish: function SUI_onSyncFinish() {
    let title = this._stringBundle.GetStringFromName("error.sync.title");

Comment 1

[Richard Newman [:rnewman]]

Are you able to reproduce this, Dietrich?

What were you doing around the time that this happened?

_stringBundle is just a lazy getter for

  return Cc["@mozilla.org/intl/stringbundle;1"].
         getService(Ci.nsIStringBundleService).
         createBundle("chrome://weave/locale/services/sync.properties");

Comment 2

[Dietrich Ayala (:dietrich)]

More info:

First there are a bunch of these in the console:

Error: gBrowser is not defined
Source File: chrome://browser/content/browser.js
Line: 3277

Which is from:

// Functions called by observers
  onActivityStart: function SUI_onActivityStart() {
    if (!gBrowser)

Then I see a bunch of these:

Error: this._stringBundle is undefined
Source File: chrome://browser/content/browser.js
Line: 3446

I was doing nothing this time. Laptop was idle. I sat down and opened the console, and this is what I saw.

Comment 3

[Philipp von Weitershausen [:philikon]]

OS X I presume? This is probably the hidden window receiving notifications. I bet we're not guarding against !gBrowser everywhere.

Comment 4

[Dietrich Ayala (:dietrich)]

(In reply to Philipp von Weitershausen [:philikon] from comment #3)
> OS X I presume? 

yessir

Comment 5

[Mike Connor [:mconnor]]

Magic 8-ball of responsibility says Ally.

Comment 6

[Nigel Babu [:nigelb]]

Attachment: patch-v1

Patch v1.  I still haven't figured how to actually get this traceback, but this should fix the wrong if(gBrowser) and if(!gBrowser) checks throughout browser-syncui.js.

Comment 7

[Gregory Szorc [:gps]]

Comment on attachment 578693 [details] [diff] [review]
patch-v1

I don't think this patch actually addresses the problem.

I believe the underlying problem is not checking for "if (!gBrowser)" in every callback. For example:

>onSyncFinish: function SUI_onSyncFinish() {
>    let title = this._stringBundle.GetStringFromName("error.sync.title");

We should bail if we don't have gBrowser defined. e.g.

>    if (!gBrowser) {
>      return;
>    }
>
>    let title = this._stringBundle...

This file is effectively concatenated to a larger file, so the GLOBAL_SCOPE workaround is not necessary.

Comment 8

[Gregory Szorc [:gps]]

My comment in the previous comment regarding the GLOBAL_SCOPE workaround isn't correct. We are seeing errors when checking gBrowser. This confounds both rnewman and myself, as gBrowser should always be in scope. No other browser-*.js file seems to conditionally check for gBrowser. We must be binding to a different scope somewhere or something.

Comment 9

[Philipp von Weitershausen [:philikon]]

(In reply to Gregory Szorc [:gps] from comment #8)
> My comment in the previous comment regarding the GLOBAL_SCOPE workaround
> isn't correct. We are seeing errors when checking gBrowser. This confounds
> both rnewman and myself, as gBrowser should always be in scope. No other
> browser-*.js file seems to conditionally check for gBrowser. We must be
> binding to a different scope somewhere or something.

This is most likely a lingering observer. The whole window has been garbage collected, including its scope, but the observer hasn't. So looking up stuff in the scope, like `gBrowser`, will fail. Happy hunting! :/