The World Series of Poker Facebook game will not start
Categories
(Web Compatibility :: Site Reports, defect, P3)
Tracking
(firefox66 unaffected, firefox67 unaffected, firefox67.0.1 unaffected, firefox68 affected)
| Tracking | Status | |
|---|---|---|
| firefox66 | --- | unaffected |
| firefox67 | --- | unaffected |
| firefox67.0.1 | --- | unaffected |
| firefox68 | --- | affected |
People
(Reporter: rdoghi, Unassigned)
References
Details
(Keywords: webcompat:needs-contact)
Attachments
(1 file)
|
203.89 KB,
image/png
|
Details |
[Affected versions]
Firefox 68.0a1
[Affected platforms]
Ubuntu 16.04, Windows 7/10, Mac OS 10.13
[Steps to reproduce]
- Reach Facebook.com and Log in.
- Reach the WSOP poker game and try to Play the game
[Expected result]
The Game should start without issues.
[Actual result]
The Cookies Disabled message is displayed.
As well as the Please ensure that your cookies are enabled and that they are not restricted in some way by your browser settings. message is also displayed.
Please note that the same issue occurs even if the user Turns of blocking for this site.
| Reporter | ||
Updated•6 years ago
|
| Reporter | ||
Comment 1•6 years ago
|
||
Hi again, I tried to do a Mozregression but there were waay to many skipped builds I couldnt grab an accurate push log,but maybe someone can guess which caused the issue here:
Comment 2•6 years ago
|
||
Please note that the same issue occurs even if the user Turns of blocking for this site.
That seems quite weird to me and indicates an error in their detection code rather than our anti-tracking being at fault. Mike, should we move this over to webcompat?
Comment 3•6 years ago
|
||
Dennis, could you investigate here?
Comment 4•6 years ago
|
||
Their cookie checks looks like this:
this.areCookiesEnabled = function() {
const e = ~~(1e3 * Math.random()) + "" + Date.now();
return (
(!navigator.hasOwnProperty("cookieEnabled") || navigator.cookieEnabled) &&
"cookie" in document &&
(document.cookie.length > 0 || (document.cookie = e).indexOf.call(document.cookie, e) > -1)
);
}
so they end up setting document.cookie to a random value like 8401557910195219. This is not supposed to work, as the value always should be a set-cookie string, so we discard it. Chrome, however, does create a new cookie with no name and a value. This has to be a bug, since this also results in the Cookie header on requests be in a wrong format, so I filed a bug.
To get this site working in all browsers, they should rewrite their test so that setting the value contains a cookie name as well, like document.cookie = "test=" + e. I don't think we should change anything on our side, unless we get feedback from Chrome people that they are not going to change their behavior.
Comment 5•6 years ago
|
||
Excellent diagnosis, thanks Dennis.
Comment 6•6 years ago
|
||
Dupe of bug 1551729?
Comment 7•6 years ago
|
||
Yes, absolutely.
Description
•