Open Bug 1651670 Opened 4 years ago Updated 2 years ago

Cut off buttons on extended share screen notification panel

Categories

(Core :: Widget: Win32, defect, P1)

All
Windows
defect

Tracking

()

Tracking Status
firefox-esr68 --- unaffected
firefox-esr78 --- wontfix
firefox78 --- wontfix
firefox79 --- wontfix
firefox80 --- wontfix
firefox81 --- wontfix
firefox82 --- fix-optional

People

(Reporter: asoncutean, Assigned: cmartin)

References

(Regression)

Details

(Keywords: regression)

Attachments

(1 file)

Attached image screenshot issue.png

Affected versions

  • 79.0a1 (2020-07-08)
  • 78.0b5
  • 78.0.1

Affected platforms

  • Windows 10
  • Ubuntu 18.04

Steps to reproduce

  1. Go to https://janus.conf.meetecho.com/screensharingtest.html
  2. Click on the Start button
  3. Insert a title and click on the Share your screen button
  4. Click on the Pick a window (or Share screen) button from the displayed popoup
  5. Select any option from the Widow or Screen to share dropdown
  6. Check the checkbox from Disable notifications from Nightly while sharing

Actual result

  • “Allow” and “Not Now” are cut off almost entirely

Expected result

  • The notification panel is extended, the “Allow” and “Not Now” buttons are not cut off

Regression range

Additional notes

  • The issue is reproducible intermittent, sometimes the checkbox needs to be toggled several times to trigger the issue
  • Reproducible with 100% scale
  • Severity: S3
Has Regression Range: --- → yes
Has STR: --- → yes

Anca, do you think you can get a narrower regression range? This looks fairly serious, but the range is too big to find out anything.

Flags: needinfo?(anca.soncutean)

I'm not able to reproduce in Windows 10, in 78.0.2 or the latest Nightly (and not able to reproduce in macOS).

Anca, does it reproduce for you in Nightly, or do we think this was somehow fixed? (if it doesn't maybe the --findfix option of mozregression is easier to work with)

(In reply to Gian-Carlo Pascutto [:gcp] from comment #1)

Anca, do you think you can get a narrower regression range? This looks fairly serious, but the range is too big to find out anything.

I can still reproduce the issue with the latest Nightly on Windows 10; see the about:support content
Note that I couldn't reproduce the issue on other two test machines with Windows 10.

Flags: needinfo?(anca.soncutean)
Severity: -- → S3

Given the regression range, seems like someone involved in widget sandboxing should investigate first.

Component: Layout → Widget: Win32
Flags: needinfo?(gpascutto)
Flags: needinfo?(cmartin)
Regressed by: 1604412
Assignee: nobody → cmartin
Flags: needinfo?(gpascutto)
Priority: -- → P1

Affected platforms
Ubuntu 18.04

Anca, the regressor you identified in comment 3 looks very plausible, but that's a Windows only change. Can you actually reproduce this on Ubuntu, or was that a typo?

If it is indeed reproducible on Ubuntu, can you try to find the regression range on that platform? It might be long-standing race that bug 1604412 made easier to hit on Windows, so if we can find the actual regressor, that could help.

Flags: needinfo?(anca.soncutean)

(In reply to Gian-Carlo Pascutto [:gcp] from comment #5)

Anca, the regressor you identified in comment 3 looks very plausible, but that's a Windows only change. Can you actually reproduce this on Ubuntu, or was that a typo?

Hmm, not really sure what happened here.. It sure looks like a mistake from my side... I rechecked the reproducible of the issue on Ubuntu with the latest builds and the one mentioned when submitting the issue and indeed, I am not able to see the described problem. Still reproducible on Windows 10 though, with the latest Nightly build.

Flags: needinfo?(anca.soncutean)
OS: All → Windows

I haven't been able to get this issue to reproduce on any of my machines.

:Anca - Is there a way for me to get access to a machine that is able to repro the issue?

Thanks!

Flags: needinfo?(cmartin) → needinfo?(anca.soncutean)

(In reply to Chris Martin [:cmartin] from comment #7)

:Anca - Is there a way for me to get access to a machine that is able to repro the issue?

Not sure the implication and politics in place for a remote access. Probably a zoom meeting with screen share is a better option. It looks like something specific to my test machine (as mentioned in comment 3 “Note that I couldn't reproduce the issue on other two test machines with Windows 10”). The issue is reproducible with a clean profile, but not reproducible in safe mode. Let me know what additional info I can provide! Screencast with the issue!

Flags: needinfo?(anca.soncutean)
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: