1408952 - A lot of clickable fields don't work in current Nightly

Status: RESOLVED DUPLICATE of bug 1408607

(Firefox :: General, defect)

Description

[Benjamin Bouvier [:bbouvier] (inactive)]

STR #1 - Fastmail.com:

- log in
- switch to a sub directory
- click an email on the middle panel

Nothing happens, while the email should be open.


STR #2 - Telegram:

- open a conversation that contains a link
- click the link

Nothing happens.


STR #3 - Gitlab:

- open an issue
- try to assign somebody to the issue

Assignable users can be hovered, but none can be selected.


I've tried without any addons, I get the same behavior.

Putting in General until I get a better target.
On my way to mozregression...

Comment 1

[Benjamin Bouvier [:bbouvier] (inactive)]

Unfortunately, the bug doesn't reproduce on Mozregression latest build (2017-10-15), so it might be a pref I have enabled.

I've checked again with all the addons disabled, and observed the same issue.

Here's my list of current interesting (as per about:support) prefs: https://benj.me/pub/prefs.txt

Comment 2

[Benjamin Bouvier [:bbouvier] (inactive)]

Disabling e10s makes the bug disappear in all 3 cases.

Comment 3

[David Ross]

Behaviour also seen in Twitter Direct Messages

Comment 4

[lilydjwg]

My Telegram doesn't work correctly too. With 2017-10-13 clicking on the conversation list doesn't do anything if I remember correctly.

With 2017-10-15 (the one I'm using right now) the conversation list works but left clicks on messages on the right only changes mouse cursor to default in somewhat random way. Selecting works. Middle and right clicks work too. Things don't change when I "open in new non-e10s window" from the context menu of the tab except that the first click works.

I get similar issues with Twitter Direct Messages list too.

Comment 5

[Benjamin Bouvier [:bbouvier] (inactive)]

(In reply to lilydjwg from comment #4)
> Things don't change when I "open in new non-e10s window" from the
> context menu of the tab except that the first click works.

Confirmed; I have to disable e10s at the global level (settings) to have it work. Opening a non-e10s window doesn't fix it.

I also confirm it was working at least on Thursday last week (12th). Not sure about the state on Friday. On Saturday it wasn't working, for sure.

Still trying to chase the pref that could cause this...

Comment 6

[lilydjwg]

I just downgrade to the first 2017-10-13 nightly (by file time on the file listing here: https://ftp.mozilla.org/pub/firefox/nightly/latest-mozilla-central/) and it works fine. So it's between the two nightlies in 2017-10-13.

Comment 7

[Pascal Chevrel:pascalc]

Using Twitter DM as the regression for mozregression here is the regression window:

Narrowed inbound regression window from [0160c57b, d6fdc1d3] (3 builds) to [bd0ad3db, d6fdc1d3] (2 builds) (~1 steps left)
51:32.55 INFO: No more inbound revisions, bisection finished.
51:32.55 INFO: Last good revision: bd0ad3dbf357a122d780925cc0981f3d4deb259a
51:32.55 INFO: First bad revision: d6fdc1d3b07044a6cd84adcc433d3f6a943dca20
51:32.55 INFO: Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=bd0ad3dbf357a122d780925cc0981f3d4deb259a&tochange=d6fdc1d3b07044a6cd84adcc433d3f6a943dca20

Comment 8

[Pascal Chevrel:pascalc]

Timothy, it seems that your patch in Bug 1405397 broke several major sites, is your follow up patch in Bug 1408607 covering these problems as well?

Comment 9

[Benjamin Bouvier [:bbouvier] (inactive)]

(In reply to Pascal Chevrel:pascalc from comment #7)

Confirming the regression window using mozregression with a clone of my current profile.

Comment 10

[Sylvestre Ledru [:Sylvestre]]

Just like pascal said.

Comment 11

[Benjamin Bouvier [:bbouvier] (inactive)]

(In reply to Pascal Chevrel:pascalc from comment #8)
> Timothy, it seems that your patch in Bug 1405397 broke several major sites,
> is your follow up patch in Bug 1408607 covering these problems as well?

Fwiw, my three issues can't be reproduced with the try build from bug 1408607 comment 1, so looks like a dup.

Comment 12

[Timothy Nikkel (:tnikkel)]

(In reply to Pascal Chevrel:pascalc from comment #8)
> Timothy, it seems that your patch in Bug 1405397 broke several major sites,
> is your follow up patch in Bug 1408607 covering these problems as well?

It should!

Comment 13

[Benjamin Bouvier [:bbouvier] (inactive)]

Confirming that the nightly with patches from bug 1408607 doesn't reproduce the issue. Thanks!

Comment 14

[Gerry Chang [:gchang]]

Tracking 58- and mark 58 fixed as bug 1408607 was fixed in 58.