1756083 - Zoom link won't open with Firefox snap

Status: RESOLVED FIXED

(Core :: Widget: Gtk, defect, P1)

Description

[William Lachance (:wlach)]

Attachment: about:support

Try to open a Zoom link with the latest Firefox snap (e.g. https://us02web.zoom.us/j/<id>)

Expected: Firefox prompts to open a URL handler, can launch Zoom
Actual: Firefox doesn't prompt to open anything, can't launch Zoom

This was working fine for me a couple of days ago so it must have regressed recently (It could also be due to an environment change, though I'm on Ubuntu 21.10 and haven't done anything other than normal updates).

Comment 1

[William Lachance (:wlach)]

Would be nice to use mozregression to track this down, but it doesn't support snaps :(

Comment 2

[William Lachance (:wlach)]

Oops, forgot to mention (though it's in about:support), this is with 98.0b6

Comment 3

[William Lachance (:wlach)]

Confirmed by installing snaps that this doesn't happen with release but does happen since 98.0b1. So presumably this bug got introduced at some point in the Nightly cycle.

Comment 4

[William Lachance (:wlach)]

Pascal, should this be tracked for the 98 release? It's a pretty significant regression on Ubuntu.

Comment 5

[Pascal Chevrel:pascalc (PTO until Sept 2)]

(In reply to William Lachance (:wlach) from comment #4)

Pascal, should this be tracked for the 98 release? It's a pretty significant regression on Ubuntu.

Definitely, thanks for the ping!
Olivier, Mike, could you have a look please? Thanks

Comment 6

[Alexandru Trif, Desktop Test Engineering [:atrif] [PTO 15-29 Aug, please ni? :mboldan]]

Hello!
Confirming as well that the issue does not reproduce with Firefox snap 97.0.1. Reproduced the issue with Firefox snap 98.0b1 and it's reproducible with Firefox snap 98.0b7 on Ubuntu 21. Same behavior on Ubuntu 18 and 20 with Firefox snap 98.0b7.

I cannot reproduce the issue with normal Firefox 98.0b7. The Firefox prompt to open the URL handler is displayed accordingly while Zoom is installed.

Unfortunately, I am not aware of a way to bisect snap builds on Nightly, if there is one please let me know.

Comment 7

[Olivier Tilloy]

I can confirm the regression. There is indeed no convenient way of bisecting snap builds, I'll do it manually to try and nail down when the regression happened.

Comment 8

[Olivier Tilloy]

Last known good revision of the snap: 901 (https://github.com/oSoMoN/firefox-snap-experiments/actions/runs/1737662092), built from https://hg.mozilla.org/mozilla-central/rev/5c51b325c09f22a0d9384cfc5198f27ec9bdbfc8.

First bad revision of the snap: 905 (https://github.com/oSoMoN/firefox-snap-experiments/actions/runs/1740004895), built from https://hg.mozilla.org/mozilla-central/rev/e40a136dc8760f3fdd8a29fad47581d16d646c80.

Likely culprit: https://hg.mozilla.org/mozilla-central/rev/1985d591b13389a5b8983393359bcc0b70084ed3

Comment 9

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

Set release status flags based on info from the regressing bug 1516290

Comment 10

[Pascal Chevrel:pascalc (PTO until Sept 2)]

Robert, your fix in bug 1516290 seems to have caused this regression. The regression is more serious than the bug it fixed so my thinking is that given that we are in our last week of beta and that the other issue was a longstanding issue, backing out bug 1516290 from beta seems reasonable. What do you think? Are there side effects we should expect from a backout? Thanks

Comment 11

[Robert Mader [:rmader]]

(In reply to Pascal Chevrel:pascalc from comment #10)

Robert, your fix in bug 1516290 seems to have caused this regression. The regression is more serious than the bug it fixed so my thinking is that given that we are in our last week of beta and that the other issue was a longstanding issue, backing out bug 1516290 from beta seems reasonable. What do you think? Are there side effects we should expect from a backout? Thanks

That seems fine to me.

Should be relatively easy to fix for nightly though (leaving ni open for that).

Comment 12

[Pascal Chevrel:pascalc (PTO until Sept 2)]

Backed out changeset 1985d591b133 (Bug 1516290) for causing a functional regression in snap builds (Bug 1756083) a=backout
https://hg.mozilla.org/releases/mozilla-beta/rev/9e37658e45e4

Comment 13

[Olivier Tilloy]

Robert, from a quick glance it looks like it would be worth making this IsRunningUnderFlatpak() function more generic, to reflect whether the application is running under confinement (flatpak, snap, or possibly others). What do you think?

Comment 14

[Robert Mader [:rmader]]

(In reply to Olivier Tilloy from comment #13)

Robert, from a quick glance it looks like it would be worth making this IsRunningUnderFlatpak() function more generic, to reflect whether the application is running under confinement (flatpak, snap, or possibly others). What do you think?

Yes, that sounds sensible to me. Unfortunately I have very limited time for that myself in the moment. Emilio, the actual author of that function, might have, however.

Comment 15

[Olivier Tilloy]

I can confirm that the regression is gone in the latest snap beta build (98.0b8-1, revision 1022). Thanks Pascal!

Comment 16

[Alexandru Trif, Desktop Test Engineering [:atrif] [PTO 15-29 Aug, please ni? :mboldan]]

We have also verified the issue with Firefox snap 98.0b8 on Ubuntu 16, 18, 20, and 21. The Firefox prompt to open the URL handler is displayed accordingly when visiting a Zoom meeting link.

Comment 17

[Emilio Cobos Álvarez (:emilio)]

Attachment: Bug 1756083 - Cleanup snap name handling and enable portal when running under either flatpak or snap. r=stransky,rmader

Comment 18

[Emilio Cobos Álvarez (:emilio)]

Attachment: Bug 1756083 - Remove useless ifdef. r=stransky,rmader

We never compile nsGNOMEShellService without MOZ_WIDGET_GTK.

Depends on D139474

Comment 19

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/800e87c42d1d
Cleanup snap name handling and enable portal when running under either flatpak or snap. r=stransky
https://hg.mozilla.org/integration/autoland/rev/8159d4ecc9f8
Remove useless ifdef. r=stransky

Comment 20

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/90fd28271d6c
Fix silly typo.

Comment 21

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/150a7e471401
Don't run snap tests in non-gtk platforms.

Comment 22

[Marian-Vasile Laza]

https://hg.mozilla.org/mozilla-central/rev/800e87c42d1d
https://hg.mozilla.org/mozilla-central/rev/8159d4ecc9f8
https://hg.mozilla.org/mozilla-central/rev/90fd28271d6c
https://hg.mozilla.org/mozilla-central/rev/150a7e471401

Comment 23

[Olivier Tilloy]

Emilio, thanks for your work on this. I filed bug #1757209 for a regression I just noticed, which happens when $SNAP_INSTANCE_NAME ≠ MOZ_APP_NAME (e.g. $SNAP_INSTANCE_NAME = firefox_nightly whereas MOZ_APP_NAME = firefox).

The implementation of GetSnapInstanceName() needs to be reverted to get the value of $SNAP_INSTANCE_NAME.

Comment 24

[Emilio Cobos Álvarez (:emilio)]

Thanks, will fix.