295813 - forbes.com - content loading into wrong iframes

Status: NEW

(Core :: DOM: Navigation, defect)

Description

[chris hofmann]

If you go to any forbes story (e.g.
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html) and refresh the page
several times, you will start to see ads loading into the wrong iframes

 

Most of the ads on the page are embedded iframes (because most commercial ad
servers’ tags use iframes, that’s the way we get them from the agencies)…the
weird behavior is that the browser seems to mix up the iframes, and load the
iframe’d pages into the wrong iframe.

 

If you reload a bunch of times soon you’ll see for example a 300x120 ad loading
into the 250x250 iframe, etc.

 

I have been doing all my testing on Firefox 1.0, but our QA team has reported
reproducing it consistently on 1.0.3 as well

Comment 1

[chris hofmann]

I think I've been able to observe this on the trunk.

seems to happen most often with when the ad for dollar rent-a-car add gets place
on the right, next to the "e-mail alerts" content, instead of on top of it.

could be something in the content management system and content or could be
surfacing a layout bug...

Comment 2

[chris hofmann]

here is the dollar rental car image that gets "squished" next to the e-mail alerts
http://spe.atdmt.com/b/FQFKQDRACDRC/SBA_336X280.gif


there are ads where it looks like the 160x600 demensions seem intended
http://m3.doubleclick.net/704727/paperairplane_160x600.gif

Comment 3

[Bob Clary [:bc] (inactive)]

Attachment: screenshot forbes-01.jpg

Chris, is this what you are seeing? That image looked designed to be that size

http://att.atdmt.com/b/FQFKQDRACDRC/SBA_120x600.gif

Comment 4

[chris hofmann]

bob,

yes, https://bugzilla.mozilla.org/attachment.cgi?id=184760 is the general layout
that I see, but when the problem crops up a wider image gets inserted in the
content http://spe.atdmt.com/b/FQFKQDRACDRC/SBA_336X280.gif 

Comment 5

[chris hofmann]

Attachment: a "narrow image" like the fedex add gets placed in a wide container

also seems to be cases where the reverse happens.   a "narrow image" like the
fedex add gets placed in a wide container

screen shot attached

Comment 6

[Bob Clary [:bc] (inactive)]

Attachment: screenshot forbes-02.jpg

screenshot of problem

Comment 7

[Bob Clary [:bc] (inactive)]

Firefox 1.0.4/Venkman crashes when trying to step through the morass of ad code
on this site and Firefox 1.1+/Deer Park has disabled the latest version of
Venkman so I can't use trunk to debug this page. I'll try with Seamonkey, but
basic functionality like Venkman needs to be kept current when we bump the
version requirements for extensions.

Comment 8

[Bob Clary [:bc] (inactive)]

Ok, it appears to occur when the adcontroller url looks like:

<http://ads.forbes.com/RealMedia/ads/adstream_mjx.ads/forbes.com/smallbusiness/story/id2056230/1220937029@Block,BigEditPromoWind,AdController,BigBanner,CenterBanner,LeftTop,RightTop,RightBottom,x5,ZebusCompanion,x50,LeftBottom,LeftBottom2,x1,x2,SponsorLogo,x70,MiddleBanner,RightLower,x33,Loge,StoryLogo,LuxuryLogo,x39,x57,LeftBottom3,LeftBottom4?adTerms=Guitars+Recording+Industry+Music+Entrepreneur+Small+Business&tickerTerms=IBM+GE+PHG>

Note the "Block"

It does not appear to occur when the adcontroller url looks like:

<http://ads.forbes.com/RealMedia/ads/adstream_mjx.ads/forbes.com/smallbusiness/story/id2056230/1118381361@RightAboveCol,RightMiddle,AdController,BigBanner,CenterBanner,LeftTop,RightTop,RightBottom,x5,ZebusCompanion,x50,LeftBottom,LeftBottom2,x1,x2,SponsorLogo,x70,MiddleBanner,RightLower,x33,Loge,StoryLogo,LuxuryLogo,x39,x57,LeftBottom3,LeftBottom4?adTerms=Guitars+Recording+Industry+Music+Entrepreneur+Small+Business&tickerTerms=IBM+GE+PHG>

Note the "RightAboveCol"

I noticed that when reloading the page when the error does not occur I see only
one ASSERT:

###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973

but when the error occurs, the ASSERT appears multiple times (for each iframe?).

The "RightAboveCol" script writes out a linked image while the "Block" ad writes
out an IFRAME

function OAS_RICH(position) {
if (position == 'Block') {
document.write ('<iframe
src="http://view.atdmt.com/FKQ/iview/frbscdrc0010000288fkq/direct/01&1061033829?click=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1061033829/Block/DollarRAC_win_SmBus_Spon_041001/DollarRAC_win_SmBus.html/34323532303934393432393862613830?"
frameborder="0" scrolling="no" marginheight="0" marginwidth="0" topmargin="0"
leftmargin="0" width="336" height="280">\n');
document.write ('<script language="JavaScript" type="text/javascript">\n');
document.write ('document.write(');
document.write ("'");
document.write ('<a
href="http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1061033829/Block/DollarRAC_win_SmBus_Spon_041001/DollarRAC_win_SmBus.html/34323532303934393432393862613830?1061033829"
target="_blank"><img border="0"
src="http://view.atdmt.com/FKQ/view/frbscdrc0010000288fkq/direct/01/1061033829"
/></a>');
document.write ("'");
document.write (');\n');
document.write ('</script>\n');
document.write ('<noscript>\n');
document.write ('<a
href="http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1061033829/Block/DollarRAC_win_SmBus_Spon_041001/DollarRAC_win_SmBus.html/34323532303934393432393862613830?1061033829"
target="_blank"><img border="0"
src="http://view.atdmt.com/FKQ/view/frbscdrc0010000288fkq/direct/01/1061033829"
/></a>\n');
document.write ('</noscript>\n');
document.write ('</iframe>\n');
}

Comment 9

[Bob Clary [:bc] (inactive)]

Looks like bug 292950 may be related

Comment 10

[Bob Clary [:bc] (inactive)]

I turned on DEBUG_PAGE_CACHE in nsDocShell.cpp and reloaded until the wrong ad
is present. The repeated errors for
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html may correspond to the
"flashing" which appears as the page is loaded.

###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html
...
Error loading URL http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html :
804b0002
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.18;sz=728x90;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1424672825/BigBanner/Fedex_bb_smbiz_050407/fedex_smbiz_bb.html/34323532303934393432393862613830?;ord=1424672825?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.19;sz=300x250;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1575753211/Block/fedex_win_smbiz_050407/fedex_smbiz_win.html/34323532303934393432393862613830?;ord=1575753211?
...
Error loading URL http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html :
804b0002
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.18;sz=728x90;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1424672825/BigBanner/Fedex_bb_smbiz_050407/fedex_smbiz_bb.html/34323532303934393432393862613830?;ord=1424672825?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.19;sz=300x250;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1575753211/Block/fedex_win_smbiz_050407/fedex_smbiz_win.html/34323532303934393432393862613830?;ord=1575753211?
Error loading URL http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html :
804b0002
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.18;sz=728x90;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1424672825/BigBanner/Fedex_bb_smbiz_050407/fedex_smbiz_bb.html/34323532303934393432393862613830?;ord=1424672825?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.19;sz=300x250;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1575753211/Block/fedex_win_smbiz_050407/fedex_smbiz_win.html/34323532303934393432393862613830?;ord=1575753211?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://cache.directorym.com/creative/module/TypeA/module.aspx?af=1076&cs=68&ss=8&b=true&source=ABTEST_NEW
...
Error loading URL http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html :
804b0002
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.18;sz=728x90;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1424672825/BigBanner/Fedex_bb_smbiz_050407/fedex_smbiz_bb.html/34323532303934393432393862613830?;ord=1424672825?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://ad.doubleclick.net/adi/N1851.forbes.com/B1573717.19;sz=300x250;click0=http://ads.forbes.com/RealMedia/ads/click_lx.ads/forbes.com/smallbusiness/story/id2056230/1575753211/Block/fedex_win_smbiz_050407/fedex_smbiz_win.html/34323532303934393432393862613830?;ord=1575753211?
...
###!!! ASSERTION: RestorePresentation should only be called for history loads:
'mLoadType & LOAD_CMD_HISTORY', file
c:/work/mozilla/anonymous/trunk/mozilla/docshell/base/nsDocShell.cpp, line 4973
no saved presentation for uri:
http://cache.directorym.com/creative/module/TypeA/module.aspx?af=1076&cs=68&ss=8&b=true&source=ABTEST_NEW
...
Document http://www.forbes.com/2005/05/26/cz_ds_0526harmonix.html loaded
successfully

over to DocShell.

Comment 11

[chris hofmann]

lets try and get this for firefox 1.1

Comment 12

[Boris Zbarsky [:bzbarsky]]

This is happening with 1.0.x builds, isn't it?  If so, it has nothing to do with
bug 292950, and might or might not have something to do with docshell...

In any case, unless you have bfcache turned _on_, bug 292950 has no visible
manifestation in non-debug builds.

Comment 13

[Bob Clary [:bc] (inactive)]

(In reply to comment #12)

happens in 1.0.x and trunk builds and even in debug, I don't have bfcache turned on.

Comment 14

[Bob Clary [:bc] (inactive)]

Just tested with a 2005-06-09 deer park on winxpsp2 and can not reproduce the
problem with the different sized ads appearing in the wrong iframes even after
reloading the page more than a dozen times. Chris, can you still reproduce that?
I do see multiple "flashes" where the entire page blanks and then redraws as the
page is loading though. The "flashing" is definitely a pita.

Comment 15

[Asa Dotzler [:asa]]

minusing unless we can demonstrate that we still have a problem here. Not
fastback related so I'm removing the dependency.

Comment 16

[Bob Clary [:bc] (inactive)]

Asa, I can reproduce this within a couple of reloads with a build from today:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050614
Firefox/1.0+

Comment 17

[Mikko Rantalainen]

Attachment: Broken screenshot while testing with Build ID 2005061903

I can reproduce this bug quite easily with build id 2005061903
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b2) Gecko/20050619
It's easiest to reproduce by first clicking refresh and trying to time the
second refresh click at the time the "flash" comes or a bit earlier. (The
"flash" is the flicker to white during the page rendering.)

I guess this has something to do with hitting refresh while some resources for
the page has been loaded but not all.

Comment 18

[Benjamin Smedberg]

per bclary this is serious enough to need fixing

Comment 19

[Bob Clary [:bc] (inactive)]

This is actually a pretty old bug. I just reproduced it in Mozilla 1.4, so it is
not a recent regression.

Comment 20

[Boris Zbarsky [:bzbarsky]]

If it's not a recent regression, then a minimal testcase that shows the problem
reliably would really help.

Comment 21

[Boris Zbarsky [:bzbarsky]]

So what could be going on here is that we always reload the "current" URI in an
iframe.  Our definition of "current" is that we have a linear list of history
entries corresponding to subframes in document order.  When doing a reload, we
load the corresponding entry from the list in each subframe.  See bug 299553
comment 1 for why we have this behavior, what IE does differently, and my
thoughts on whether (and how, if so) we should do what IE does.

In any case, our reload behavior means that if the DOM changes significantly
between reloads we could get an effect like that described in this bug.  In
particular, say the original DOM has:

  <iframe src="foo" style="height: 100px; width: 300px;"></iframe>
  <iframe src="bar" style="height: 300px; width: 100px;"></iframe>

And after the reload the DOM looks like:

  <iframe src="bar" style="height: 300px; width: 100px;"></iframe>
  <iframe src="foo" style="height: 100px; width: 300px;"></iframe>

When we reload this page, we will end up with the "foo" page loaded in the first
(tall and skinny) iframe, and the "bar" page in the second (short and wide)
iframe, which sounds a lot like the behavior being described in this bug.

Comment 22

[RNicoletto]

I cannot reproduce:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b3) Gecko/20050714
Firefox/1.0+ (djeter)

Windows XP Pro SP2

Comment 23

[Chris]

i no longer see this problem on win2000, Mozilla/5.0 (Windows; U; Windows NT
5.0; en-US; rv:1.8b4) Gecko/20050806 Firefox/1.0+ ID:2005080619

Comment 24

[Brendan Eich [:brendan]]

Removing blocking-aviary1.5 flag.

/be

Comment 25

[Bob Clary [:bc] (inactive)]

Chofmann, can you still reproduce this? I just reloaded the page a dozen or more
times with switches between the wide and narrow ad formats with FF15b2/winxpsp2
and did not see the problem.

Comment 26

[chris hofmann]

yeah,  seems to be working ok now -- using 1.5 b2

Comment 27

[vosv9md02]

Have observed this same problem using Firefox 2.x, including 2.0.0.3 on WinXP. So, still not fixed.

Comment 28

[Michal Matula]

Yes, the problem still persist  with 2.0.0.11, even with the nightly build of 3.x from 12-04-2007.
I have added a test case files and comment to bug # 315891 (which seemed to be related before I found this one)
using XP

Comment 29

[treeform]

Still a problem, it looks like this bug is famous - i feel these bugs are some how related:

https://bugzilla.mozilla.org/show_bug.cgi?id=276824
https://bugzilla.mozilla.org/show_bug.cgi?id=301325
https://bugzilla.mozilla.org/show_bug.cgi?id=315891
https://bugzilla.mozilla.org/show_bug.cgi?id=342905
https://bugzilla.mozilla.org/show_bug.cgi?id=363840

Comment 30

[Brett Stime]

Probable dup of bug 279048

Comment 31

[Sebastian Nohn]

Attachment: Test Case

Testcase. Can be executed locally. You may have to reload multiple times. Please also note the instructions on top of the HTML file See isb_ok.png for expected result, isb_nok.png for actual resulty.

Comment 32

[Sebastian Nohn]

Attachment: Test Case: Expected Result

Comment 33

[Sebastian Nohn]

Attachment: Test Case: Actual Result

Comment 34

[Laura Thomson :laura]

bsmedberg suggested getting a reproducible test case on this bug in order to get some traction toward fixing it.  ccing him since we now have one.  (Thanks!)

(I understand Sebastian is also interested in sponsoring someone to work on a fix, if this bug isn't on our radar.)

Comment 35

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

I am aware of at least one scenario in which something like this is plausible.

If multiple subframe loads race and a forward navigation loses, that will screw up the session history tree and could possibly result in things ending up in the wrong docshells.

See Bug 669110 comment 2.

Comment 36

[Simona Badau, Desktop QA]

Mozilla/5.0 (Windows NT 5.1; rv:26.0) Gecko/20100101 Firefox/26.0
Build ID: 20130806030203

I couldn't reproduce the initial issue (tried using the testcase provided in Comment 31) on the latest Nightly nor on Firefox 4.0. 

Removing the qawanted keyword since the test case that was requested in Comment 20 was provided in Comment 31. Please re-add it back if anything else is needed from the QA side.

Comment 37

[Stephen Donner [:stephend] Not actively reading bugmail]

Mass-removing myself from cc; search for 12b9dfe4-ece3-40dc-8d23-60e179f64ac1 or any reasonable part thereof, to mass-delete these notifications (and sorry!)