Bug 75554 – REGRESSION (r104000): AdBlock extension fails to load/function

• Component: DOM
• Status: RESOLVED FIXED

[Fabian Mailinator]

it thinks it's loaded, but it's not.


to do:
install adblock for safari 2.5.11 in r104022 nightly
visit any page on the internet.

notice:
ads!

expected:
no ads!

[Alexey Proskuryakov]

Adam, looks like this broke in <http://trac.webkit.org/changeset/103999> or <http://trac.webkit.org/changeset/104000>.

[Alexey Proskuryakov]

<rdar://problem/10644493>

[Andy Estes]

Confirmed this is due to <http://trac.webkit.org/changeset/104000> via a local build. It looks like AdBlock is failing to inject its <style> element into the DOM after r104000.

[Adam Barth]

Is there a test case I can try?

[Andy Estes]

(In reply to comment #4)
> Is there a test case I can try?

No, this needs a reduction.

[Fabian Mailinator]

Test case;


1. Install adblock for safari: 
    a. open safari.
    b. menu: Safari -> Safari Extensions ...
    c. Adblock for safari is currently the most popular extension
        and at the top of the list.
    d. click on it and follow installation instructions.


2. Test adblock for safari:
    a. visit http://google.com
    b. RIGHTclick on any non-active area of the page.
    c. If adblock is functioning, the pop-up menu will
       include the item "Block any add on this page"
    d. If adblock is not functioning, the pop-up menu 
       will NOT include a "block" item.

[Adam Barth]

> No, this needs a reduction.

Andy, are you planing to create a reduction, or should I start from the repro instructions in Comment #6?

[Alexey Proskuryakov]

*** Bug 75613 has been marked as a duplicate of this bug. ***

[Andy Estes]

(In reply to comment #7)
> > No, this needs a reduction.
> 
> Andy, are you planing to create a reduction, or should I start from the repro instructions in Comment #6?

I'm focused on something else right now and wouldn't be able to look at this for the next few days. I'd recommend starting with the repro steps for now.

[Joe Strzemp]

If you want to reproduce this issue with AdBlock,
you have to use a version of AdBlock PRIOR to 2.5.12.

Taken from the AdBlock Changelog:

2.5.12: Refactoring, and work around a bug in Chrome 18 that breaks AdBlock (famlam)"

[Andy Estes]

Here is the AdBlock ticket that tracks this issue: <http://code.google.com/p/adblockforchrome/issues/detail?id=6360>

Here is the Chromium bug filed by an AdBlock engineer: <http://code.google.com/p/chromium/issues/detail?id=109272>

Looks like the information in the AdBlock ticket would be sufficient for creating a reduction, but I've only skimmed it.

[Andy Estes]

The issue would appear to be that, prior to r104000, document.documentElement would be non-null for scripts injected at document start. After r104000, document.documentElement is null at document start.

[Adam Barth]

Thanks!  That should be easy to fix.

[Adam Barth]

Created attachment 122964 [details]
work in progress (needs tests)

[Adam Barth]

Created attachment 122966 [details]
Patch

[WebKit Review Bot]

Comment on attachment 122966 [details]
Patch

Clearing flags on attachment: 122966

Committed r105321: <http://trac.webkit.org/changeset/105321>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.