The behavior reported here is possibly open for debate. But since the current behavior is not what our other mobile and desktop browsers do, I am defaulting to a bug.
Steps to reproduce:
Expected:
Actual:
Cc to @jonalmeida because I think he implemented Private Mode in the Reference Browser, which has this same bug.
I'd say it's definitely a bug for the reason you mentioned: not consistent with mobile and desktop.
That said, this sounds like a GV bug since we're only setting a flag to change a session to private mode, and I can reproduce this on Focus as well.
Just confirmed that this was closed by #667.
I can still reproduce this issue in PB. You are also logged out in the first tab if you refresh the page after step 3.
I agree with @jonalmeida: This seems to be an "issue" at GeckoView level. I filed a bug on Bugzilla for that:
https://bugzilla.mozilla.org/show_bug.cgi?id=1533406
Thanks for filing that, @pocmo đ
@bifleming this looks like it's in the Q4 waiting column but it was added in June, so I propose we bump it out of that list, and changing it to a P3.
Need to decide what the expected behaviour is for this one. If different from current behaviour, we'll need to prioritize with GV
Since @AmyYLee has been working on privacy and security projects, letâs wait for her feedback.
In the meantime, I have an opinion that may be useful:
I think it makes sense that private tabs should share cookies â especially for tabs that are spawned from a âparentâ tab. This allows internal links on Facebook, for example, to work properly.
Whether this would be considered a âMustâ, âCouldâ or âShouldâ, I tend towards the latter.
This is my reasoning:
Interestingly, what seems to happen is that a subset of technical, privacy-conscious users (granted, this was on desktop) seem to prefer each private tab _not_ to share cookies.
On Fennec, cookies are shared between private tabs. After migration, users may consider this as a regression.
Also, some websites open new tabs and as pointed by sv-ohorvath https://github.com/mozilla-mobile/fenix/issues/609#issuecomment-470430757, when a second tab is opened, the first tab is also logged out.
Since @AmyYLee has been working on privacy and security projects, letâs wait for her feedback.
In the meantime, I have an opinion that may be useful:
I think it makes sense that private tabs should share cookies â especially for tabs that are spawned from a âparentâ tab. This allows internal links on Facebook, for example, to work properly.
Whether this would be considered a âMustâ, âCouldâ or âShouldâ, I tend towards the latter.
This is my reasoning:
* Weâve designed Private Browsing to be erasable in one tap (garbage bin) and to self-destruct on quit. * Therefore, itâs reasonable to assume that users are less likely to keep private tabs open, and therefore signed in. * Therefore, it doesnât matter very much whether information on one private tab carries over to another private tab.Interestingly, what seems to happen is that a subset of technical, privacy-conscious users (granted, this was on desktop) seem to prefer each private tab _not_ to share cookies.
* https://bugzilla.mozilla.org/show_bug.cgi?id=117222 * https://bugs.chromium.org/p/chromium/issues/detail?id=15632 * https://bugs.chromium.org/p/chromium/issues/detail?id=24690
Hi,
I agree with @brampitoyo points for sharing cookies across private tabs and we want to be consistent with desktop. If we do hear users feeling strongly about not sharing cookies in private tabs, this should be considered across all browsers (ie desktop and mobile) and we should keep it consistent no matter what we choose to do. In this case we should share cookies across private tabs. Thanks!
Thanks @AmyYLee ! Yes, that sounds good to me. That's the current behaviour on iOS as well and it makes sense if you're researching and opening multiple related tabs
Bug 1533406 is scheduled to be worked on in December.
This seems fixed on nightly for me
Sounds like this should be fixed. QA please verify :)
Cookies are kept until the end of the private session: