1207700 - Use XInput 2 device manager on GTK3

Status: REOPENED

(Core :: Widget: Gtk, defect, P2)

Description

[Andrew Comminos [:acomminos]]

We had to disable XI2/multidevice support in bug 1170342 due to focus and scrolling related issues. We should monitor these issues to see if it would be possible to re-enable XI2 in the future.

Comment 1

[Kuba Niewiarowski]

Please add option to reenable via about:config or something - I'm really got used to it...

Comment 2

[Martin Stránský [:stransky] (ni? me)]

(In reply to Kuba Niewiarowski from comment #1)
> Please add option to reenable via about:config or something - I'm really got
> used to it...

Just set MOZ_USE_XINPUT2=1 env variable which enables it in Firefox.

Comment 3

[Kuba Niewiarowski]

(In reply to Martin Stránský from comment #2)
> Just set MOZ_USE_XINPUT2=1 env variable which enables it in Firefox.

Not working...
I'm using Aurora builds from https://launchpad.net/~ubuntu-mozilla-daily/+archive/ubuntu/firefox-aurora

Comment 4

[Anders Kaseorg]

MOZ_USE_XINPUT2=1 is in Nightly (bug 1207973) but not yet in Aurora.

Comment 6

[sgl4kn]

Using "MOZ_USE_XINPUT2=1" &e10s worked for me (https://bugzilla.mozilla.org/show_bug.cgi?id=1268599#c10), but not very smooth (scrolling, afterwords opening the menu item, then trying to scroll results in an un-scrollable page). Fedora already compiles FF with "MOZ_USE_XINPUT2=1", which is very smooth on my tablet.

However, it would be interesting how this is handled by Firefox for Android, which clearly needs touch support and XINPUT2.

Comment 8

[Martin Stránský [:stransky] (ni? me)]

I think we can enable the XInput2 for recent Gtk+ versions (say Gtk+ >= 3.20) as the related Gtk+ focus bugs was fixed but for XInput2 only. Also Fedora enables XInput2 by default and we don't have any user complains.

Comment 9

[Martin Stránský [:stransky] (ni? me)]

Before the switch we need to address Bug 1182700. Bug 1170342, Bug 1196777 should be fixed for reasonable recent systems.

Comment 10

[Martin Stránský [:stransky] (ni? me)]

For Bug 1182700 looks like gnome folks gave up (https://bugzilla.gnome.org/show_bug.cgi?id=750994#c8) and recommend to use Wayland. I wonder how big complication is it and if benefits from enabled xinput2 outweigh it, especially when the focus bugs should be fixed on recent systems.

Comment 11

[Koala Yeung]

Gnome issue 750994 has been migrated to their Gitlab instance:
https://gitlab.gnome.org/GNOME/gtk/issues/558

Comment 13

[QwertyChouskie]

Is there any possibility of enabling this soon? Out-of-the-box Firefox is unusable with touchscreens, and as touchscreens become more common (e.g. I just upgraded my laptop to one that has a touchscreen), having this enabled quickly becomes quite important for a large number of users. Also, https://bugzilla.mozilla.org/show_bug.cgi?id=1182700 was just fixed in Kwin, and AFAIK that was the last blocker of this bug.

Also, AFAIK, Fedora enables MOZ_USE_XINPUT2 by default for these reasons, so apparently it's not too buggy :P

Comment 14

[QwertyChouskie]

After some more thinking, it would probably make sense to enable it by default but disable it for KWin < 1.19.3, so those users don't get hit by the bug.

Comment 15

[Botond Ballo [:botond]]

(In reply to QwertyChouskie from comment #13)

Also, https://bugzilla.mozilla.org/show_bug.cgi?id=1182700 was just fixed in Kwin, and AFAIK that was the last blocker of this bug.

Based on comment 10, it sounds like Gnome is affected as well, and the issue is not fixed there?

Comment 16

[QwertyChouskie]

I am unable to reproduce using the steps in https://bugzilla.gnome.org/show_bug.cgi?id=750994#c0 with my laptop's touchpad, tested on Gnome Shell 3.36 (Ubuntu 20.04). I tried the same steps with FF and also can not reproduce it there. I suspect the bug was mitigated/had its impact lessened (perhaps unintentionally) in Gnome at some point, which then left the issue for KDE users (as per most of the comments in https://bugzilla.mozilla.org/show_bug.cgi?id=1182700).

Comment 17

[QwertyChouskie]

(Also FYI, the KWin fix was just backported to the 5.18 series, so it should be fixed in the next 5.18.x point release, meaning Ubuntu 20.04 and its derivatives should get this fix soon.)

Comment 18

[Botond Ballo [:botond]]

Martin, WDYT? Based on the above info, should we enable XInput 2 for KWin >= 5.19.3 (and >= 5.18.next)? How about other window managers?

Comment 19

[Martin Stránský [:stransky] (ni? me)]

(In reply to Botond Ballo [:botond] from comment #18)

Martin, WDYT? Based on the above info, should we enable XInput 2 for KWin >= 5.19.3 (and >= 5.18.next)? How about other window managers?

Fedora has used XINPUT2 by default for five years and I don't have any reports so I guess it's okay to enable it for recent systems, at least for KDE/Gnome.

Comment 20

[Botond Ballo [:botond]]

(In reply to Martin Stránský [:stransky] from comment #19)

Fedora has used XINPUT2 by default for five years and I don't have any reports so I guess it's okay to enable it for recent systems, at least for KDE/Gnome.

Thanks!

(As this bug is assigned to you, I assume you'll write the enablement patch, but please let me know if you'd like me to.)

Comment 21

[Martin Stránský [:stransky] (ni? me)]

Attachment: Bug 1207700 [Linux] Enable XInput2 on Gtk 3.24 and newer, r?botond

Comment 22

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/6c7f2177e485
[Linux] Enable XInput2 on Gtk 3.24 and newer, r=botond

Comment 23

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/6c7f2177e485

Comment 24

[Martin Stránský [:stransky] (ni? me)]

Let's revert that in Bug 1660212.

Comment 25

[Mathew Hodson]

This bug should probably be reopened since the change was completely reverted.

Comment 26

[pierpaolo.franco]

(In reply to Mathew Hodson from comment #25)

This bug should probably be reopened since the change was completely reverted.

Confirmed. Need to set environment variable in Debian 11 with Firefox 78.12.0esr (64 bit).

Comment 27

[BugBot [:suhaib / :marco/ :calixte]]

The severity field for this bug is relatively low, S3. However, the bug has 3 duplicates, 22 votes and 50 CCs.
:stransky, could you consider increasing the bug severity?

For more information, please visit auto_nag documentation.

Comment 28

[BugBot (nomail) [:suhaib / :marco/ :calixte]]

The last needinfo from me was triggered in error by recent activity on the bug. I'm clearing the needinfo since this is a very old bug and I don't know if it's still relevant.