1249083 - Google Sheets not evaluating cells

Status: RESOLVED DUPLICATE of bug 1257861

(Core :: Networking: Cookies, defect)

Description

[Stephanie Hobson [:shobson]]

Google Sheets is not evaluating new formulas when I enter them. This makes Google Sheets unusable.

I know this is Google's fault but I also know we have to clean up after them when they break things in Firefox. I don't know if this bug is of interest to you to fix or not. 

Steps to reproduce:
1) create new sheet
2) in A:1 type "1"
3) in B:1 type "2"
4) in C:1 type "=sum(A1:B1)"
5) progress meter appears in editing bar (top right of screen below tool bars)
6) nothing appears in C:1

Expected result:
C:1 should output 3.

Comment 1

[:Gijs (he/him)]

This works for me. What version of Firefox are you using? Have you retried on a clean profile? (https://support.mozilla.org/en-US/kb/profile-manager-create-and-remove-firefox-profiles )

Comment 2

[Stephanie Hobson [:shobson]]

Reported in 45.0b6 but it does not happen on a clean profile.

Comment 3

[chris hofmann]

wondering what could possible be in a profile that could cause the content of a spreadsheet to be incorrect.  does it still happen in your profile and 45.0.1 release?

Comment 4

[Stephanie Hobson [:shobson]]

I haven't blown away my profile yet. It's still happening to me in 46.0b1.

Comment 5

[chris hofmann]

maybe the first place to look for would be to see if there are any non-standard pref settings for some experimental features, debugging modes, or other things in js, networking, dom or other gecko areas.

Comment 6

[chris hofmann]

Another approach on comment 5 would be to see if we can get someone to run the simple test case in the description that exercises the sum() function on a google spreadsheet page, look at the code that's running in the browser, and see if there are any runtime settings around that code that could affect this working correctly.

one more thing to look at would be to check for particular addons that might be installed (like ad blockers) that might be changing the way that javascript interacts with pages.  Can you list out the addons installed?

miket might have some contacts at google that could tell us if there are any circumstances or settings that might cause the simple sum() test case to fail on the google spreadsheet content side.

this is the kind of bug that could be catastrophic if it would ever be exposed to more than a handful of people, or in certain kinds of mission critical situations or applications; so understanding more about it has some value even if it turns out that Stephanie is the only person in the world that is likely to do something to her profile that could cause a simple sum() function to not work.  At least that would all help us to rest easier at night knowing how its possible for this to break.

Comment 7

[chris hofmann]

also, if there is a real bug here its more likely to be in gecko than the Firefox front end.  I'm going to kick it over there and open up for just a bit to see if anyone else on the gecko side has ideas.  this may bounce again to addons, or evangelism, or....

Comment 8

[chris hofmann]

:shobson,  one more thing to try on your profile where this is broken would be to trace back on 45.0b5, 45.0b4, 45.0b3, etc. to see if you can determine if its broken on all older releases, or if it stopped working at some point.  from there we might be able to associate to some gecko code change.

Comment 9

[Alice0775 White]

I can reproduce the problem on Firefox45.0.1 if third-party cookies were blocked.

Comment 10

[Alice0775 White]

(In reply to Alice0775 White from comment #9)
> I can reproduce the problem on Firefox45.0.1 if third-party cookies were
> blocked.

Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=97419358e5899b3e6ee2e21e7d89106d90b8fa70&tochange=bc44c9e55ef0e042c5a6c88bd241c08282b27f55

Regressed by: Bug 1171215

Comment 11

[Mike Taylor [:miketaylr]]

Yes, we have good contacts with folks at Google (and Google Docs), but as Alice0775 pointed out -- it's our regression. :(

Comment 12

[Stephanie Hobson [:shobson]]

I run a few things that could block 3rd party cookies (Ghostory, uBlockOrigin, Privacy Badger) I narrowed the problem down to Privacy Badger. I'm a little confused since all the cookies are from google.com but maybe subdomains are treated as 3rd parties by Privacy Badger?

I reported the problem to Privacy Badger.

I thought I'd tested with addons disabled before reporting the bug, sorry for the rookie error.

Comment 13

[chris hofmann]

great job tracking this down!  I have to admit I did lose a few winks over this bug last night, and some weird cookie thing in the profile was one of the things that came to mind.

do we have a sense from telemetry how many users might block 3rd party cookies and might be exposed to this bug?

Comment 14

[Alice0775 White]

[Tracking Requested - why for this release]: it is worth track esr45

I think the offending patch(bug 1171215) should be backed out from esr45.

Comment 15

[Blake Kaplan (:mrbkap) (inactive)]

This is probably a dupe of bug 1257861. We should recheck this when that bug is fixed.

Comment 16

[chris hofmann]

Once a fix is available I really think we should consider chemspilling for this on 45.  

The combination of this bug plus maybe fixing

Bug 1257861 - Firefox 45 fails to send Cookie header with XHR post requests done from a web worker when third-party cookies are blocked

and 

Bug 1257236 - The "inbox.google.com" page keeps refreshing on and on with "Accept third-party cookies: Never" checked

leads to some pretty serious affects for anyone blocking 3rd party cookies (privacy advocates and others that are passionate users of Firefox).  Privacy badger alone claims 72k users off amo, and more might be installing directly off the eff site https://www.eff.org/privacybadger

Comment 17

[chris hofmann]

[Tracking Requested - why for this release]:

see comment 16

Comment 18

[Honza Bambas (:mayhemer)]

Blake, can I duplicate this bug to to bug 1257861?

Comment 19

[Blake Kaplan (:mrbkap) (inactive)]

This works with my patch for bug 1257861.

Comment 20

[chris hofmann]

Wed and Thursday when I was testing this with 45.0.1 it was pretty easy to reproduce.

1) make a spreadsheet like described in the description
2) see the result of the sum() function in c1
3) disable 3rd party cookies like we advise on sumo
    https://support.mozilla.org/en-US/kb/disable-third-party-cookies
4) shift reload the spreadsheet page
5) see the result in cell c1 now blank

now when I try that I see the correct result of the sum() in cell c3.

I wonder if google has may a config change to make sheets not use third party cookies now, or if some how I just don't have enough coffee yet this morning to test right.

:shobson and  Alice0775 White :  do you see things working correctly now in your profiles that surfaced the bug and with the sheets you have been using to spot the bug and test?

Comment 21

[Alice0775 White]

I can STIL reproduce the problem with STR comment #0 on latest Nightly.

https://hg.mozilla.org/mozilla-central/rev/8a4359ad909fe0cffcfea512770483ffdf8cd4e6
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:48.0) Gecko/20100101 Firefox/48.0 ID:20160326030430

Comment 22

[chris hofmann]

Attachment: Screen Shot 2016-03-26 at 7.41.22 PM.png

haven't quite figure it out but it seems I'm getting the different behavior depending on when I've created the =sum() cell and I've had 
1) some combination of turning on/off link sharing v. private on the spreadsheet
2) some combination of flipping accept 3rd party cookies between never and always

It also could be there is some caching in the content or the cookie setting that's tripping me up.   Attachment shows 3 lines that I've copied and pasted the same cell contents as I was rolling though one of the sharing states and one of the 3rd party cookie setting states. The second line is broken and the other two lines are working after one of page reloads.

Comment 23

[Mike Conley (:mconley) (:⚙️)]

ni'ing mrbkap, re: comment 21.

Comment 24

[Blake Kaplan (:mrbkap) (inactive)]

bug 1257861 isn't checked in yet.

Comment 25

[Sylvestre Ledru [:Sylvestre]]

Tracked in the other bug.