920468 - Inconsistent tick box behaviour with cookie preferences

Status: NEW

(Firefox :: Security, defect)

Description

[sjekuopz]

If "accept third party cookies" is unticked and "accept cookies from sites" is ticked, if "accept cookies from sites" is unticked and reticked it also enables the other option.

This behaviour doesn't follow the usual behaviours for sub-options in FF and in other software - that if greyed-out when something triggers an ungreying-out the option doesn't change too.

The tickbox "always use private browsing mode" for example greys-out other options, but does not fiddle with the actual settings of the personal security options.

See also the General tab, downloads. The 2nd box there doesn't change option when the first box greys it out or not.

No matter the reasoning behind this odd functionality on the 3rd party cookie option, when the world's biggest advertising company is Mozilla's main funder, this concerns me. In fact, I don't even think IE pulls a privacy-threatening trick like this!

Comment 1

[Mihai Morar, (:MihaiMorar)]

I can confirm this is happening on Windows 7 too ( the accept third party cookies issue)

> See also the General tab, downloads. The 2nd box there doesn't change option when the first box greys 
> it out or not.

This is expected to happen, it is not a bug because If you check/uncheck "Show the Downloads window when downloading a file" option, the sub-option called "Close it when all downloads are finished" keeps the status you had used last time.

Marking as new due to the third party cookies.

Comment 2

[:Gijs (he/him)]

This seems to have been a conscious decision made by the people working on an internal preference storage change back in Firefox 1.5 (!). See bug 419596. I already discussed some reasons why I think we may want to keep the current behaviour with Mihai on IRC (briefly: because turning off cookies altogether breaks significant parts of the web, and it may be argued that turning off 3rd party cookies completely does the same, albeit to a smaller extent. So let's say a user did this, found out lots of stuff broke, and turned it back on, you want to make sure to turn it on "properly". In other words, re-enabling this checkbox automagically is an "anti-footgun", as it were, although I understand that it's weird that toggling the checkbox twice gives the result that it does. We could add more checks for that, but I'm not sure that's worth the added code complexity for a scenario that few users will run across (ie, toggling it twice without ever closing the dialog)).

Note that on nightly branches, we automatically switch to "From Visited" rather than "Always" if you toggle the main pref, and AFAIK we're still talking about what exactly we want to do with cookies to begin with.

I'll let Gavin chime in here about what the most sensible thing should be, although I suppose a minor change that'd be less invasive and probably non-controversial *might* be not changing it if it were already "From Visited" (so if it was previously "never", you'd still get "Always" to un-footgun users, but if it were previously "From Visited", we would leave it at that. But that's pretty magical, so... I'm not convinced either way).

Comment 3

[:Gijs (he/him)]

(In reply to :Gijs Kruitbosch from comment #2)
> Firefox 1.5

Err, I meant Fx 3. But still, old!

Comment 4

[Mihai Morar, (:MihaiMorar)]

(In reply to :Gijs Kruitbosch from comment #2)

Thanks Gijs for information, I had posted the comment before we talked on IRC. I'm CC-ing Tanvy mabye she can help us in any way. I worked a lot with her on Security features and she has more experience that I have.

Comment 5

[:Gavin Sharp [email: [e-mail]]]

Sorry for letting this request linger for so long. I'm inclined to think the  current behavior is "fine", in that we probably don't want to simplify further and lose the "anti-footgun", and we also probably don't want to significantly complicate this code further to address this oddity or make the tweak from comment 2.

Comment 6

[Tanvi Vyas[:tanvi]]

If you again (In reply to :Gijs Kruitbosch from comment #2)
> We could add more checks for that, but I'm
> not sure that's worth the added code complexity for a scenario that few
> users will run across (ie, toggling it twice without ever closing the
> dialog)).
> 
What do you mean by without every closing the dialog?  This also happens when you close the preference pain and reopen it (i.e. open preferences, turn off cookies, close preferences, open preferences and re-enable cookies.  The third party cookie setting is changed from "Never" to "From Visited" on Nightly).

(In reply to :Gavin Sharp [email: [e-mail]] from comment #5)
> Sorry for letting this request linger for so long. I'm inclined to think the
> current behavior is "fine", in that we probably don't want to simplify
> further and lose the "anti-footgun", and we also probably don't want to
> significantly complicate this code further to address this oddity or make
> the tweak from comment 2.

Is this really an anti-footgun?  If a user turns of cookies completely, they are purposefully (or accidentally) breaking the web more than it may already be broken from their third party cookie settings.  Then re-enabling cookies just sets them back to the state they were at before.  Why are we changing the third party cookie setting in that case?  Do we think that users with third party cookies accidentally turned off will turn off cookies to try and figure out why their browser is broken?