813623 - window.URL.createObjectURL is not a function when going through Xrays (Xrays for static properties don't work)

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[rmlipman]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:18.0) Gecko/18.0 Firefox/18.0
Build ID: 20121119042013

Steps to reproduce:

I tried to use this script in Nightly with Scriptish: http://ms11.github.com/4chanSoundPlayer/

On 4chan there are threads called "sound threads" where you post images with sounds embedded in them. In the post body there are words in [brackets]. Clicking on the words in brackets plays the embedded sounds.

The latest known working Nightly build is 2012-11-08. I last tested on 2012-11-20. Aurora has not yet been updated to 19, but since it was broken in Nightly 19 I suspect it might break in Aurora as well.


Actual results:

Instead of the sounds playing, I get this error in the web console: "TypeError: window.URL.createObjectURL is not a function". I had the script log values and at the runtime of the script window.URL is defined, but window.URL.createObjectURL has a value of undefined.

If I check the values on the web console by hand, window.URL and window.URL.createObjectURL are both defined.



Expected results:

The sound should have played.

The easiest way to test this is to go to http://catalog.neet.tv/v/ or http://catalog.neet.tv/a/ and search for "sound" or "[" to find an active sound thread.

Comment 1

[Alice0775 White]

Suspected: Bug 810520

Comment 2

[Boris Zbarsky [:bzbarsky]]

So is this only a problem in a Scriptish user script?

This should have been resolved in bug 778152, in general....

Comment 3

[Boris Zbarsky [:bzbarsky]]

And a small script I tried over here that just does this:

// ==UserScript==
// @name        Test script
// @namespace   bzbzbz
// @include     http://www.mozilla.org/*
// @version     1
// ==/UserScript==
alert(window.URL)
alert(window.URL.createObjectURL)

works fine for me.  I do see the problem with the original script, though... So something is fishy.

Can someone try reducing the script to the smallest one that shows the problem?

Comment 4

[Alice0775 White]

I am not sure the following  is correct behavior of this bug.

Str 
1. set devtools.chrome.enabled to true
2. Open a web page
3. Open ScratchPad(Shift+F4) and turn on "Browser" of environment menu
4. Paste alert(content.URL.createObjectURL)
5. Execute > run (Ctrl+R)

Actual results:
  undefined

Expected results:
  function createObjectURL() {
    [native code]
  }

Comment 5

[Alice0775 White]

(In reply to Boris Zbarsky (:bz) from comment #3)
> And a small script I tried over here that just does this:
> 
> // ==UserScript==
> // @name        Test script
> // @namespace   bzbzbz
> // @include     http://www.mozilla.org/*
> // @version     1
> // ==/UserScript==
> alert(window.URL)
> alert(window.URL.createObjectURL)
> 
> works fine for me.  I do see the problem with the original script, though...
> So something is fishy.
> 
> Can someone try reducing the script to the smallest one that shows the
> problem?

I can repro with Scriptish 0.1.8 and the small script．
Second alert shows "undefined".

http://hg.mozilla.org/releases/mozilla-aurora/rev/5f19747d3410
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:19.0) Gecko/19.0 Firefox/19.0 ID:20121120042014

Comment 6

[Boris Zbarsky [:bzbarsky]]

Using the steps in comment 5 in my debug build I get the expected result, not undefined...

Same thing in a 2012-11-19 Mac nightly.  :(

Something is definitely very fishy...

Comment 7

[Boris Zbarsky [:bzbarsky]]

Aha!  Now I've reproduced it in a debug build.  Had to make sure about:config wasn't the thing that was still showing in the browser.  ;)

Comment 8

[Boris Zbarsky [:bzbarsky]]

Looks like we never actually init the ids in sStaticMethods_ids.  I would have thought we had tests for this, but apparently not.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Attachment: Fix Xrays to actually work correctly for static methods and properties.

Comment 10

[Boris Zbarsky [:bzbarsky]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/448bb73bddb0

Comment 11

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 683743 [details] [diff] [review]
Fix Xrays to actually work correctly for static methods and properties.

[Approval Request Comment]
Bug caused by (feature/regressing bug #): Bug 792675 
User impact if declined: Extensions and user scripts can't access certain DOM
   methods that they used to be able to access.
Testing completed (on m-c, etc.): Passes added automated tests, testcases in this
   bug.
Risk to taking this patch (and alternatives if risky): Low risk.  Just makes us
   run the same code for static methods and attributes as we already run for all
   sort of other methods/attributes.
String or UUID changes made by this patch: None.

Comment 12

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/448bb73bddb0

Comment 13

[Alex Keybl [:akeybl]]

Comment on attachment 683743 [details] [diff] [review]
Fix Xrays to actually work correctly for static methods and properties.

[Triage Comment]
Will prevent the possibility for new add-on regressions in FF19 caused by Bug 792675.

Comment 14

[Boris Zbarsky [:bzbarsky]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/d112badf7960