Unexpected ABE behaviour since FF 16

[Mad55]

I'm using NoScript 2.5.7 and this ABE rules in USER to limit facebooks datamining:

Code: Select all

Site facebook.com *.facebook.com facebook.net *.facebook.net fbcdn.com *.fbcdn.com  fbcdn.net *.fbcdn.net
Accept from *.facebook.com *.facebook.net *.fbcdn.com *.fbcdn.net facebook.com
Anon

In Firefox 15.0.1 I could type in "https://www.facebook.com" in addressbar and was logged in into facebook. Now after updating to Firefox 16 I end up on the welcome landing page. If i press F5 or "reload page"-button, I'm logged in again. If I re-enter the address, I still end up on welcome landing page.

If I disable NoScript, I can type in "https://www.facebook.com" and I'm automatically logged in.

I can verify this behaviour in Aurora build 17.0a2 also.

[Thrawn]

I'm using NoScript 2.5.7 and this ABE rules in USER to limit facebooks datamining:

Code: Select all

Site facebook.com *.facebook.com facebook.net *.facebook.net fbcdn.com *.fbcdn.com  fbcdn.net *.fbcdn.net
Accept from *.facebook.com *.facebook.net *.fbcdn.com *.fbcdn.net facebook.com
Anon

In Firefox 15.0.1 I could type in "https://www.facebook.com" in addressbar and was logged in into facebook. Now after updating to Firefox 16 I end up on the welcome landing page. If i press F5 or "reload page"-button, I'm logged in again. If I re-enter the address, I still end up on welcome landing page.

If I disable NoScript, I can type in "https://www.facebook.com" and I'm automatically logged in.

I can verify this behaviour in Aurora build 17.0a2 also.

First things first: you can use a leading dot to cover both the domain and its subdomains. So, your rule could collapse to:

Code: Select all

Site .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Accept from .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Anon

Secondly, I can't reproduce this; if I'm logged in to Facebook, then I can type in the address and I'll still be logged in.

However, since ABE logs all activity to the Error Console (Ctrl + Shift + J or Tools - Web Developer - Error Console), can you look in the Info section and post any related messages? Most likely you need to add another domain to your rule.

[Mad55]

I modified ABE rule like you said and found following output in error console on accessing facebook via address:

Code: Select all

[ABE] <.facebook.com .facebook.net .fbcdn.com .fbcdn.net> Anonymize on {GET https://www.facebook.com/ <<< moz-nullprincipal:{dacad0e7-7c2a-4dee-874c-f791d9d64374} - 6}
USER rule:
Site .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Accept from .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Anonymize

If i reload the page, no new ABE message appears.

[Giorgio Maone]

It seems they started assigning unique origins to URLs opened from the address bar (previously the origin was chrome:something).

Code: Select all

Site .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Accept from .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Anon INC

Should work (it drops cookies from inclusion requests, e.g. "Like" buttons).

[Thrawn]

This looks a lot like my Gmail problem.

[Mad55]

It seems they started assigning unique origins to URLs opened from the address bar (previously the origin was chrome:something).

Code: Select all

Site .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Accept from .facebook.com .facebook.net .fbcdn.com .fbcdn.net
Anon INC

Should work (it drops cookies from inclusion requests, e.g. "Like" buttons).

This works perfectly!

[F-3000]

Code: Select all

[ABE] <.facebook.com .facebook.net .fbcdn.com .fbcdn.net> Anonymize on {GET https://www.facebook.com/ <<< moz-nullprincipal:{dacad0e7-7c2a-4dee-874c-f791d9d64374} - 6}

I keep getting similar block with everything I have ABE rule with, when trying to enter the site.

Code: Select all

Pyyntö {GET https://www.facebook.com/ <<< moz-nullprincipal:{be16eb15-7026-4ae9-97f2-47334b0ec6d9} - 6} suodattanut ABE: <.facebook.com .fbcdn.net .facebook.net .ea2d.com> Deny

Code: Select all

Site .facebook.com .fbcdn.net .facebook.net .ea2d.com
Accept from .facebook.com .fbcdn.net .ea2.com .facebook.net mindjolt.com .satumaa.eu
Deny

[Thrawn]

I keep getting similar block with everything I have ABE rule with, when trying to enter the site.
<snip>

Code: Select all

Site .facebook.com .fbcdn.net .facebook.net .ea2d.com
Accept from .facebook.com .fbcdn.net .ea2.com .facebook.net mindjolt.com .satumaa.eu
Deny

Hmm...you could change it to Deny INC, as suggested earlier, instead of just Deny, but this is cropping up more and more. Looks like it will need a more thorough workaround.

What about changing it to something that will match all web traffic, but not the special moz-nullprincipal origin? Anyone have any comments on this?

Code: Select all

Site ...
Accept from ...
Deny from ^https?://.* ^about:.*

[Giorgio Maone]

What about changing it to something that will match all web traffic, but not the special moz-nullprincipal origin? Anyone have any comments on this?

I'm working to identify browser "initial" loads (such as those from the URL bar, from bookmarks and from the search box) which were previously shown as originating from chrome and now have unique origins.
As soon as I can do it (if it's possible) I'm gonna treat them as "same origin", as it was before.

[F-3000]

As soon as I can do it (if it's possible) I'm gonna treat them as "same origin", as it was before.

I hope it'll be possible. Until then, I'll just turn the USER-rules off.

[GµårÐïåñ]

I love how mozilla keeps screwing with this. How many times now they have flip flopped on this?

[Thrawn]

I'm working to identify browser "initial" loads (such as those from the URL bar, from bookmarks and from the search box) which were previously shown as originating from chrome and now have unique origins.
As soon as I can do it (if it's possible) I'm gonna treat them as "same origin", as it was before.

That sounds good for fixing the old behavior. Even if possible, though, I don't think it will fix the Gmail manifestation of this issue (linked above), because that opens a new page (with moz-nullprincipal origin) in response to clicking on a page element. I've encountered it when opening email attachments with Google Docs, and also when switching to a delegated account.

I can PM my Gmail account details if that would help investigate it.

[Thrawn]

Update: My ABE rule is now preventing me from logging into Gmail at all. I've had to disable the rule.

Code: Select all

[ABE] <.google.com .google.com.au .youtube.com> Anonymize on {GET https://accounts.google.com/ServiceLogin?service=mail&passive=true&rm=false&continue=https://mail.google.com/mail/?shva%3D1&ss=1&scc=1&ltmpl=default&ltmplcache=2#inbox <<< https://mail.google.com/mail/?shva=1#inbox, moz-nullprincipal:{4fc66ebb-f80f-4b84-911e-f58b6c4c2d28} - 6}
USER rule:
Site .google.com .google.com.au .youtube.com
Accept from .google.com .google.com.au .youtube.com
Accept from ^moz-nullprincipal:\{.*}
Anonymize

[F-3000]

Code: Select all

Accept from .google.com .google.com.au .youtube.com
Accept from ^moz-nullprincipal:\{.*}

Does this really work? I mean both, two Accept-lines, and the moz-nullprincipal?

[Giorgio Maone]

Should be fixed in latest development build 2.5.8rc2, thank you.