463639 - cnn US presidential elections page never finishes loading (CORS request + redirect back to same-origin)

Status: RESOLVED WONTFIX

(Web Compatibility :: Site Reports, defect, P5)

Description

[Christopher Blizzard (:blizzard)]

This URL:

http://www.cnn.com/ELECTION/2008/results/president/

Has results loaded after the page is loaded.  In trunk version:

Build identifier: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b2pre) Gecko/20081107 Minefield/3.1b2pre

It never finishes loading.

Works in Firefox 3.0.3 on the mac.

Comment 1

[Radek 'sysKin' Czyz]

On error console I'm getting:

Error: illegal character
Source File: http://i.cdn.turner.com/cnn/z/.e/js/2.0/election/packages/president_static-511.js
Line: 467, Column: 3

but I'm not sure if it's related (I don't have 3.0.3 to check if the exception is thrown there as well).

Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1b2pre) Gecko/20081107 Minefield/3.1b2pre

Comment 2

[Christopher Blizzard (:blizzard)]

Interesting.  I don't get any errors on the error console at all.  And there doesn't appear to be anything obvious in the file at that line.

Comment 3

[Radek 'sysKin' Czyz]

Do you have javascript.options.showInConsole = true? Otherwise nothing shows.

That line 467 has an eval() statement and I suppose invalid character is in the variable.

The variable is:
([{\k\e\y:"us\e\r\_\t\o\p\i\c\s",\v\a\l\u\e:{\v\i\s\i\t\s:15,\t\w\_\a\r\r\a\y:[{\k\e\y:"\A\i\r\_\D\i\s\a\s\t\e\r\s",\w\e\i\g\h\t:2.2696356501000006},{\k\e\y:"\W\o\r\l\d\_\P\o\l\i\t\i\c\s",\w\e\i\g\h\t:1.5112310850900004},{\k\e\y:"\A\c\c\i\d\e\n\t\s\_\a\n\

Green arrow under the line points to first backslash (4th character).

Comment 4

[Christopher Blizzard (:blizzard)]

I do have that set, yes.  Weird.

Comment 5

[Gabriel Chadwick]

Still getting it with Mozilla/5.0 (Windows; U; Windows NT 6.0; fr; rv:1.9.1b2pre) Gecko/20081124 Minefield/3.1b2pre

Comment 7

[Ria Klaassen (not reading all bugmail)]

Regression range is http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2008-08-30+19%3A00%3A00&enddate=2008-08-31+13%3A00%3A00
so this seems to be caused by bug 407216.

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I think this should block.

Comment 9

[Mike Beltzner [:beltzner, not reading bugmail]]

We can't reproduce this anymore - perhaps fixed by another TraceMonkey fix? Please re-open if you can reproduce on latest branch or trunk nightly.

Comment 10

[Ria Klaassen (not reading all bugmail)]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090128 Minefield/3.2a1pre

The movie still doesn't load here with the latest hourly.

Comment 11

[Mike Beltzner [:beltzner, not reading bugmail]]

Ria: can you hunt down a regression range on this for us? Might go back a bit.

Comment 12

[Jeff Muizelaar [:jrmuizel]]

I can't reproduce this on Vista with:
Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.2a1pre) Gecko/20090203 Minefield/3.2a1pre (.NET CLR 3.5.30729)

Comment 13

[Ria Klaassen (not reading all bugmail)]

Attachment: Bug in action

Mysterious that some people see the flash and others don't.

Comment 14

[Tim Riley [:timr]]

WFM in B3Pre on Mac.

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1b3pre) Gecko/20090210 Shiretoko/3.1b3pre

Also WFM on:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.6) Gecko/2009011913 Firefox/3.0.6
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b3pre) Gecko/20090210 Shiretoko/3.1b3pre

Maybe another fix corrected this or the page was changed.  

Do we still need qawanted and regressionwindow-wanted?

Comment 15

[Radek 'sysKin' Czyz]

Still fails for me, still the same javascript esxception thrown as explained in comment 3
I volunteer to find regression window but in ~6 hours. If someone's faster, go ahead.

Comment 16

[Radek 'sysKin' Czyz]

Alright, I was bored so I did that sooner.
Works: 20080930093007 http://hg.mozilla.org/mozilla-central/rev/99cd24b71102
Fails: 20081001 http://hg.mozilla.org/mozilla-central/rev/63b0791327ab

Before you point out the most obvious, no, the change in UA string length was not the cause. I checked by faking the "working" UA string.

There exists a major obvious difference when tracing http headers.

When it works:
firefox asks www.cnn.com:
    GET /ELECTION/2008/json/US/national.html? HTTP/1.1
www.cnn.com responds:
    HTTP/1.x 302 Found
    Location: http://edition.cnn.com/ELECTION/2008/json/US/national.html?
firefox asks edition.cnn.com:
    GET /ELECTION/2008/json/US/national.html? HTTP/1.1
edition.cnn.com responds:
    HTTP/1.x 200 OK

So, normal get - redirect - get - success.

When it fails:
firefox asks www.cnn.com:
    OPTIONS /ELECTION/2008/json/US/national.html HTTP/1.1
www.cnn.com responds:
    HTTP/1.x 405 Method Not Allowed

.... so the very first "ask" is for OPTIONS instead of GET. Also note lack of question mark at the end of URL. Moreover, json seems to be involved. 

Finally, the javascript exception form comment 3 was something specific to my profile, please ignore it.

Comment 17

[Radek 'sysKin' Czyz]

http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2008-09-30+09%3A30&enddate=2008-10-01+05%3A00

Lots of changes... there's no hourly archive from back then is there.

Comment 18

[Radek 'sysKin' Czyz]

http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2008-09-30+07%3A30&enddate=2008-10-01+05%3A00

Better list because it includes both changesets. Apologies for the bugspam.

Comment 19

[Carsten Book [:Tomcat]]

i can reproduce the problem with  Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.9.1b3pre) Gecko/20090212 Shiretoko/3.1b3pre and latest flash version - i don't see the results and it keeps loading - also no errors in the error console.

Comment 20

[Tim Riley [:timr]]

There is now a regression range (thanks Radek!) and a few more examples  of which configurations pass and fail (thanks tomcat!).  I am removing qawanted and regressionwindow-wanted.  Feel free to add qawanted bac on if there is anything else we can help with!

Comment 21

[Andreas Gal :gal]

Do we still have a working testcase for this? I can't reproduce with TM tip, but I couldn't reproduce before either. I am tempted to close this again WFM. Objections? Can anyone still reproduce this?

Comment 22

[Carsten Book [:Tomcat]]

note, i also see the problem in  Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.0.6) Gecko/2009011912 Firefox/3.0.6 - page never finishes loading !

Comment 23

[Matthias Versen [:Matti]]

The flash doesn't load for me in a current SM branch nighly and also with Firefox 3.0.6 (!) with flash 10 U22. I'm sure that this worked before and it still works in IE. tomcat got the same issue with FF3.0.6

Comment 24

[Samuel Sidler (old account; do not CC)]

I'm going to mark this wanted for 1.9.0.x since it's blocking 1.9.1, but we won't push on it to get fixed or mark it blocking until there's a clear reproducible case (seems to go back and forth) and/or a fix ready for m-c and 1.9.1.

Comment 25

[Jeff Walden [:Waldo]]

Taking and investigating...

Comment 26

[Jeff Walden [:Waldo]]

I can't reproduce this either.

Comment 27

[Jeff Walden [:Waldo]]

Still can't reproduce.  If someone else can, please reopen -- but do note that I don't think a bug in a page of such temporal utility should block in any case.  :-)

Comment 28

[Ria Klaassen (not reading all bugmail)]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090421 Minefield/3.6a1pre

I can still reproduce it easily. Maybe location related? But I doubt indeed if it is useful to reopen.

Comment 29

[Frosch]

Bug still reproducible here, too. I think it should remain open until there is at least a definite explanation for this bug and, more importantly, if it is due to the site or to firefox. The latter would undoubtly require fixing. Just imagine something like this would happen again on such a popular site soon after the next major update to firefox is released...

Comment 30

[Tim Riley [:timr]]

Re-opening as it is clearly reproducible.

Frosch- Please give your user agent string like Ria did so we know exactly what platform and build you are using.

5 People have seen this and/or repro'ed it and 3 can't repro it.  We have a regression range (c7, c16)

Comment 31

[Frosch]

My user agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2a1pre) Gecko/20090421 Minefield/3.6a1pre

Btw., I have another Minefield running on Windows Vista at work and the bug occurs there as well.

Comment 32

[Jeff Walden [:Waldo]]

Moving back to nominated status, I don't actually think this should block per comment 27...

The location-based wrinkle might be relevant here, but I'm not sure of a way for me to verify that theory.  :-\

Comment 33

[Brendan Eich [:brendan]]

Anyone tried reducing a saved version of the site content to minimize the testcase? How about eyeballing the code, as this could be a DOM bug, not a JS bug.

/be

Comment 34

[Ria Klaassen (not reading all bugmail)]

Yeah, ok I see now: on the top panel you need to make a choice between the CNN International Edition (in that case you don't see the flash), but with CNN US edition you can always see it. This choice is stored in a cookie.

Comment 35

[Ria Klaassen (not reading all bugmail)]

I see in post bug 407216 builds while loading the choice changes from "US" to "International". The bullet in the radio button goes to "International" while loading. And "International" does not get the flash. That seems to be the bug :/

Comment 36

[Frosch]

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2a1pre)
Gecko/20090421 Minefield/3.6a1pre

Confirmed. Changing the Edition to "CNN U.S." makes the flash work again. Changing it back "CNN International" will make it fail to load.

Comment 37

[Andreas Gal :gal]

I can reproduce the bug now.

Comment 38

[Andreas Gal :gal]

This happens with JIT off as well. Could someone confirm? (about:config and disable jit for content).

Comment 39

[Andreas Gal :gal]

Works with Safari.

Comment 40

[Jeff Walden [:Waldo]]

Yeah, fails with JIT off (and Flashblock disabled for edition.cnn.com, stupid site)...

Comment 41

[Andreas Gal :gal]

Did anyone try capturing the international version of the page? That might reproduce offline.

Comment 42

[Andreas Gal :gal]

Attachment: offline testcase

Comment 43

[Andreas Gal :gal]

offline testcase doesn't work in safari either

Comment 44

[Andreas Gal :gal]

offline testcase doesn't work due to document.domain mucking which fails a security check in the offline case

Comment 45

[Andreas Gal :gal]

I can't find any indication that this is JS related. I works in 3.0.8 and Safari. We probably need a bisect to get somewhere with this.

Comment 46

[Carsten Book [:Tomcat]]

I will see that i can find a regression range for this !

- Tomcat

Comment 47

[Ria Klaassen (not reading all bugmail)]

I verified and got exactly the same result, Bug 407216. If the bug still is that it doesn't load the flash with the "International" cookie.

Comment 48

[Radek 'sysKin' Czyz]

(In reply to comment #46)
> I will see that i can find a regression range for this !

You mean something different than comment #16?

Comment 49

[klonos]

Confirming in Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2a1pre) Gecko/20090422 Minefield/3.6a1pre (.NET CLR 3.5.30729) ID:20090422044118

changing to US loads, switching to international fails

Comment 50

[Ria Klaassen (not reading all bugmail)]

In case someone would want http logs for comparison between a correctly working and a failing build, I have them here. They are large to attach.

Comment 51

[Robert Sayre]

I can get this to never finish loading on Firefox2, Firefox3.0.x, Shiretoko, and Trunk.

Comment 52

[Henrik Skupin [:whimboo][⌚️UTC+1]]

There is a problem in downloading the whole pages. Firefox states that the source file cannot be read?! Is someone able to grab all the files?

Comment 53

[Michael Ryan]

I think there are actually two issues going on here.

First, there was a regression caused by bug 407216 ("setting a property that has only a getter") which has since been fixed, though I haven't tracked down what changeset fixed it. After that regression was fixed, this bug was no longer reproducible, so it was closed.

Sometime later, a second regression popped up that was only reproducible when using the "international edition". International edition changes the domain to edition.cnn.com, so the XHR for http://www.cnn.com/ELECTION/2008/json/US/national.html is cross-site. So it seems likely that this bug in its current form is the same as bug 485704.

Moving to Tech Evangelism...

Comment 54

[Michael Ryan]

(In reply to comment #53)
> First, there was a regression caused by bug 407216 ("setting a property that
> has only a getter") which has since been fixed, though I haven't tracked down
> what changeset fixed it. After that regression was fixed, this bug was no
> longer reproducible, so it was closed.

This was fixed the day after bug 407216 landed, so this bug is really all about cross-site XHR. The 2008-08-31 regression window was a red herring.

Comment 55

[Jeff Walden [:Waldo]]

Whee!  Michael, you are awesome.

Comment 56

[Robert Sayre]

wow, nice work Michael.

Comment 57

[Karl Dubost💡 :karlcow]

As of today this is still an issue. 
There is an infinite loading wheel into Firefox Aurora version 31.

In Opera 21, we get the results.
Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/34.0.1847.76 Safari/537.36 OPR/21.0.1432.31 (Edition Next)

In Safari Version 7.0.4 (9537.76.4), we get the results.

Moving to Desktop Component.

Comment 58

[Hallvord R. M. Steen [:hallvors]]

"Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at http://www.cnn.com/ELECTION/2008/json/US/national.html. This can be fixed by moving the resource to the same domain or enabling CORS."

If you send an OPTIONS request for http://www.cnn.com/ELECTION/2008/json/US/national.html the site returns a redirect! :-o

OPTIONS http://www.cnn.com/ELECTION/2008/json/US/national.html HTTP/1.1
Host: www.cnn.com
User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:29.0) Gecko/20100101 Firefox/29.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
DNT: 1
Origin: http://edition.cnn.com
Access-Control-Request-Method: GET
Access-Control-Request-Headers: x-prototype-version,x-requested-with
Connection: keep-alive
Pragma: no-cache


HTTP/1.1 302 Moved Temporarily
Server: nginx
Date: Wed, 11 Jun 2014 09:42:43 GMT
Content-Type: text/html
Content-Length: 154
Connection: keep-alive
Set-Cookie: CG=NO:01:Nesoddtangen; path=/
Cache-Control: no-cache
Vary: *
Location: http://edition.cnn.com/ELECTION/2008/json/US/national.html

So that's why. Chrome permits a CORS request that redirects back to a same-origin URL to succeed, Firefox looks at the lack of CORS headers in the response and terminates..

Which behaviour is right per spec?

Comment 59

[Anne (:annevk)]

Firefox is correct: http://fetch.spec.whatwg.org/#cors-preflight-fetch (if status is not in the range 200 to 299 a network error is returned). Once there is a preflight in play, redirects are banned.

What does Chrome do exactly? Does it do another OPTIONS against the new URL? Where does it do the actual GET/POST against?

Comment 60

[Hallvord R. M. Steen [:hallvors]]

Ah, in Chrome it's a GET request, not OPTIONS - no CORS involved, it loads content with an IFRAME.
So the code that caused bug 485704 is indeed striking again.

Comment 61

[Karl Dubost💡 :karlcow]

Contacted someone from CNN today about it.

Comment 62

[Karl Dubost💡 :karlcow]

Recontacted the person from CNN, but the email is not working anymore

> Google tried to deliver your message, but it was rejected by the server for the recipient domain turner.com by atlmail3.turner.com. [157.166.168.181].

So I guess the personal is not working there anymore. 

We need to find a new contact.

Comment 63

[Hallvord R. M. Steen [:hallvors]]

(Since this CNN page is about an old election - however historic and exciting - we might at some point decide to give up making them fix it. But what's up with Chrome getting the IFRAME-based branch - does it still throw the exception we should not throw anymore?)

Comment 64

[Oana Arbuzov [:oanaarbuzov]]

I don't ever see CNN ever fixing this, thus I will close it as Wontfix.