Closed Bug 56545 Opened 25 years ago Closed 25 years ago

mozilla allows cookie for the whole .com.br

Categories

(Core :: Networking: Cookies, defect, P3)

x86
Linux
defect

Tracking

()

RESOLVED DUPLICATE of bug 8743

People

(Reporter: cesarb, Assigned: morse)

References

()

Details

From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.0-test9-pre1 i686; en-US; m18) Gecko/20001009 BuildID: 2000100908 Mozilla seems to allow a site to set a cookie which applies to the whole .com.br domain, instead of ignoring the cookie. Reproducible: Always Steps to Reproduce: 1. Enable confirmation of cookie requests and set it to allow from the originating server only 2. Open the page Actual Results: Mozilla asked if I wanted to set a cookie for the whole .com.br domain. If I agree, it sets a RMID cookie for the whole .com.br second level domain. Expected Results: Ignored the cookie (not even bringing up the confirmation dialog), possibly saying in some way (a popup or a funny yellow or red icon) that the site was using a invalid HTTP header. Crashing before setting the cookie might be acceptable too =)
Unfortunately there is no way that this bug can ever be fixed. See discussion in bug 8743.
Status: UNCONFIRMED → RESOLVED
Closed: 25 years ago
Resolution: --- → WONTFIX
Nope, I disagree. You can always: a) Strictly follow the "Enable cookies for originating site only" and disallow any domain-level cookies b) Creatively follow the "Enable cookies for originating site only" and allow all kind of wierd cookied, BUT store the server that sent them in the cookies database and ONLY send them to the originating site. I'm not concerned about privacy nonsense. I'm concerned about bandwidth wastage. If every freaking domain decided to do so for .com.br, a http request would take minutes. Besides, I believe legitimate sites seldom (if ever) use a cookie which should be sent to another server (if I get a cookie from bugzilla.mozilla.org, it'll probably be used again only in bugzilla.mozilla.org).
a) We can't disallow domain-level cookies. It's part of the cookie spec and many sites utilize it. b) We can't change the cookie database without losing backwards compatibility.
Aw, of course you can! Make a cooksite.txt and record THERE which cookie is from which site. You lost no backwards compatibility (since the other two files are still there, and in the same format, and with the same information as before).
OK, I stand corrected. We can modify the cookie database in the manner you described and still maintain backwards compatibility (I should have realized that since I'm the one who introduced cookperm.txt for that very reason). But what you are proposing (in b) is a change in cookie semantics and you run the risk of breaking legitimate web-sites. Many well-intentioned websites do very strange things with cookies. Take a look at all the trouble we ran into with yahoo mail when we tried to improve upon the situation (gory details given in bug 8743). In fact, from experience I would have to disagree with the following statement that you made: "I believe legitimate sites seldom (if ever) use a cookie which should be sent to another server"
OK, I understand now... At least make this a dupe of 8743 then.
Making dup as requested by reporter.
Status: RESOLVED → UNCONFIRMED
Resolution: WONTFIX → ---
*** This bug has been marked as a duplicate of 8743 ***
Status: UNCONFIRMED → RESOLVED
Closed: 25 years ago25 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.