1442528 - Keyboard handling in Google Docs is broken if "dom.keyboardevent.keypress.dispatch_non_printable_keys_only_system_group_in_content" is true

Status: RESOLVED FIXED

(Web Compatibility :: Site Reports, defect, P1)

Description

[Lars T Hansen [:lth]]

Backspace/Delete doesn't work any more.  Arrow keys don't work.  Shift+TAB doesn't work.  Looks like something is busted in our keyboard handling.

No obvious errors in the console.

Observed both on Linux and macOS with the Nightly that is most recent as of 0900 CET 2018-03-02.

Comment 1

[Chris Peterson [:cpeterson]]

Masayuki, your fix for keypress event bug 1440189 broke keyboard handling in Google Docs.

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away 4/27 - 5/6)]

Yeah, should be fixed by Google. Makoto-san has already contacted them.

Comment 3

[Lars T Hansen [:lth]]

I can confirm that the workaround is effective, thanks!

Comment 5

[Andrew Overholt [:overholt]]

Should we move this to tech evangelism?

How long should we leave this broken (in the default state) waiting for a GDocs fix?

Comment 8

[Šime Vidas]

Breaking Google Docs is not a good idea, I think. I’ve set dom.keyboardevent.keypress.dispatch_non_printable_keys_only_system_group_in_content to false to “fix” this issue, but before I knew about this (yesterday), I had to switch to Firefox stable. So, people like me who default to Nightly and use Docs daily will have an issue with this change.

Comment 9

[Preston]

Are we just waiting for Google Docs to fix this? Does turning dom.keyboardevent.keypress.dispatch_non_printable_keys_only_system_group_in_content to false have any other effects anywhere else? It does work, though. Even then, I'm getting a lot of input lag.

Comment 10

[Makoto Kato [:m_kato] (PTO until 5/6)]

(In reply to Andrew Overholt [:overholt] from comment #5)
> Should we move this to tech evangelism?

Yes.

> How long should we leave this broken (in the default state) waiting for a
> GDocs fix?

Nakano-san, could you answer this?

Comment 12

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away 4/27 - 5/6)]

(In reply to Makoto Kato [:m_kato] from comment #10)
> (In reply to Andrew Overholt [:overholt] from comment #5)
> > How long should we leave this broken (in the default state) waiting for a
> > GDocs fix?
> 
> Nakano-san, could you answer this?

As far as possible because we need to collect reports of other broken web apps on Nightly.

Note that the change was requested by Google :-(

# But of course, I'm afraid that a lot of Nightly testers changes the pref for using Google Docs and/or Gmail.

Comment 13

[Pascal Chevrel:pascalc]

(In reply to Masayuki Nakano [:masayuki] (JST, +0900) from comment #12)
> (In reply to Makoto Kato [:m_kato] from comment #10)
> > (In reply to Andrew Overholt [:overholt] from comment #5)
> > > How long should we leave this broken (in the default state) waiting for a
> > > GDocs fix?
> > 
> > Nakano-san, could you answer this?
> 
> As far as possible because we need to collect reports of other broken web
> apps on Nightly.
> 
> Note that the change was requested by Google :-(
> 
> # But of course, I'm afraid that a lot of Nightly testers changes the pref
> for using Google Docs and/or Gmail.


I think this is not a good approach. 
The most advanced users will change the pref and the more casual users will just switch back to another browser to get work done. We are breaking nightly for our users by making gdocs and gmail, two very popular web applications, unusable. The end result will be that we will loose nighty users and the telemetry and crash reports they send daily. Our nightly population is a small and finite resource, for one user that reports a broken web app because of this change, how many are leaving us?

Is there an ETA given by Google to have their web apps fixed? I would suggest a more collaborative approach with our nightly testers community by disabling the feature and writing a blog post about the change and asking for our nightly testers to flip the pref to true and report the broken web apps they find.

Comment 14

[Jim Jeffery not reading bug-mail 1/2/11]

I would like to know how Google Docs is involved in the 'Search bar' on the 'New Tab Page'.

1.  Open a new tab by clicking on the '+' 
2.  Focus the search box and enter a search item
3.  Try to arrow down to make a selection for the suggestions.
4.  Navigation is not possible from the keyboard

5.  Flip the pref noted above and now navigation works.  
dom.keyboardevent.keypress.dispatch_non_printable_keys_only_system_group_in_content

Seems to me that something else is going on in Nightly Firefox besides just Google stuff.

Tested on Latest Nightly build running on Win10 x64.

Comment 15

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away 4/27 - 5/6)]

(In reply to Pascal Chevrel:pascalc from comment #13)
> The most advanced users will change the pref and the more casual users will
> just switch back to another browser to get work done. We are breaking
> nightly for our users by making gdocs and gmail, two very popular web
> applications, unusable. The end result will be that we will loose nighty
> users and the telemetry and crash reports they send daily.

Hmm, it's reasonable reason to back it out temporarily.

> Is there an ETA given by Google to have their web apps fixed?

Not sure. I just cced the thread, and I'm not a member of the ML, so, I'm not sure if I received all emails of the thread.

Comment 16

[Andrew Overholt [:overholt]]

(In reply to Masayuki Nakano [:masayuki] (JST, +0900) from comment #15)
> (In reply to Pascal Chevrel:pascalc from comment #13)
> > The most advanced users will change the pref and the more casual users will
> > just switch back to another browser to get work done. We are breaking
> > nightly for our users by making gdocs and gmail, two very popular web
> > applications, unusable. The end result will be that we will loose nighty
> > users and the telemetry and crash reports they send daily.
> 
> Hmm, it's reasonable reason to back it out temporarily.

I think we should do that.

Comment 17

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away 4/27 - 5/6)]

(In reply to Andrew Overholt [:overholt] from comment #16)
> (In reply to Masayuki Nakano [:masayuki] (JST, +0900) from comment #15)
> > (In reply to Pascal Chevrel:pascalc from comment #13)
> > > The most advanced users will change the pref and the more casual users will
> > > just switch back to another browser to get work done. We are breaking
> > > nightly for our users by making gdocs and gmail, two very popular web
> > > applications, unusable. The end result will be that we will loose nighty
> > > users and the telemetry and crash reports they send daily.
> > 
> > Hmm, it's reasonable reason to back it out temporarily.
> 
> I think we should do that.

Already, done. See bug 1443117.

Comment 20

[reinhart_previano]

Duplicate of this bug: ~~https://webcompat.com/issues/~~

Comment 21

[reinhart_previano]

Duplicate of this bug: https://webcompat.com/issues/15831

Comment 22

[Julien Cristau [:jcristau]]

This was backed out for now.

Comment 23

[reinhart_previano]

Duplicate of this bug: https://webcompat.com/issues/15790

Comment 25

[Daniel Holbert [:dholbert]]

(In reply to Jim Jeffery not reading bug-mail 1/2/11 from comment #14)
> I would like to know how Google Docs is involved in the 'Search bar' on the
> 'New Tab Page'.

(Just to follow up on this part: this New Tab Page issue is tracked in bug 1454262.)

Comment 26

[Panos Astithas (he/him) [:past] (please ni?)]

Google has fixed this on their end and the changes should be in production mid-next week. We should remove Google domains from the dom.keyboardevent.keypress.hack.dispatch_non_printable_keys pref and do a QA run by the end of next week or the beginning of the week after that.

Comment 27

[Liz Henry (:lizzard) (relman/hg->git project)]

Panos, is there any followup still to do here as you mention in comment 26?
On a quick check to a new Google Doc, tab, arrow keys, backspace, and delete work for me.

Comment 28

[Panos Astithas (he/him) [:past] (please ni?)]

We have been in communication with the Google folks and are waiting for them to deploy the fix in production. They hit some snags and reported that were trying to resolve them more than a month ago. When we receive an updated ETA I will update this bug.

The reason it works now Liz is that we have effectively disabled our change for every domain present in the dom.keyboardevent.keypress.hack.dispatch_non_printable_keys pref.

Comment 29

[Liz Henry (:lizzard) (relman/hg->git project)]

Thanks Panos! I'll mark this disabled in nightly too

Comment 30

[Panos Astithas (he/him) [:past] (please ni?)]

We got confirmation from Google that fixes for Gmail, Docs and Keep are now in production. I have asked for clarification on Hangouts and Inbox, but we should get QA to look at this ASAP. I've been running with the google domains removed from dom.keyboardevent.keypress.hack.dispatch_non_printable_keys and it seems to be working fine.

Comment 31

[Panos Astithas (he/him) [:past] (please ni?)]

Got confirmation that Hangouts and Inbox have deployed the fix, too. Can we get QA to perform some smoke testing with these entries removed from the pref?

Comment 32

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away 4/27 - 5/6)]

Yeah, I confirmed the fix too. I'll file a new bug to remove *.google.com from the blacklist.