New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 638389 link

Starred by 1 user

Issue metadata

Status: ExternalDependency
Owner: ----
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: iOS
Pri: 3
Type: Bug



Sign in to add a comment

iOS cookie parsing differs from defacto standards

Project Member Reported by jww@chromium.org, Aug 16 2016

Issue description

Because Chrome on iOS relies on the OS cookie line parser (in particular, NSHTTPCookie and its parsing method cookiesWithResponseHeaderFields, see https://developer.apple.com/library/mac/documentation/Cocoa/Reference/Foundation/Classes/NSHTTPCookie_Class/#//apple_ref/occ/clm/NSHTTPCookie/cookiesWithResponseHeaderFields:forURL:), we cannot currently control how cookies are parsed.

NSHTTPCookie seems to be pretty much spec compliant, but that leads to some noticeable differences with the de facto standards of all the other UAs, such as with empty-key cookies. See http://inikulin.github.io/cookie-compat/#NAME0028 for an example of where iOS differs, similar to what Safari does.

Chrome is aiming to match these de facto standards, as per  issue 601786 , but we can't do that on iOS. Thus, for now, we are blocking out some tests on iOS, which would fail (such as CookieStoreTest::EmptyKeyTest), but if there comes a day where either NSHTTPCookie matches the de facto standard, or we do our own cookie parsing on iOS, we should re-enable these tests.
 
Cc: pkl@chromium.org
Labels: M-55
Owner: eugene...@chromium.org
Status: Assigned (was: Untriaged)
eugenebut@ can you address this?

Comment 2 by jww@chromium.org, Aug 17 2016

FWIW, I'm not sure there's anything to do in the short-term. I meant for this to be a tracker bug so that we have it on our plate for a future where either we aren't depending on NSHTTPCookie or its behavior is changed.
Cc: eugene...@chromium.org
Labels: -M-55
Owner: ----
Status: ExternalDependency (was: Assigned)
Joel, should we file a radar for this issue? If so, may I ask you to do this (because I'm not familiar with details of the problem)?.

Comment 4 by jww@chromium.org, Aug 17 2016

No, I don't think we should right now. Technically they're following the spec, and we're only changing Chrome's behavior to match what the majority of other UAs are doing. There's a chance we'll add this behavior to the spec at some point, in which case it makes sense to file a radar, but not right now.
Owner: jww@chromium.org
Is this still an issue?
Owner: eugene...@chromium.org
Status: Assigned (was: ExternalDependency)
Is this still an issue?
Should we close this bug?
Owner: ----
Status: ExternalDependency (was: Assigned)
Still an issue.

Sign in to add a comment