[linux] Rearranging the pinned tabs by throwing them around pinned tabs might overlap and break the UI
Categories
(Core :: Widget: Gtk, defect, P3)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox-esr140 | --- | unaffected |
| firefox140 | --- | unaffected |
| firefox141 | --- | unaffected |
| firefox142 | --- | affected |
| firefox143 | --- | affected |
People
(Reporter: rdoghi, Unassigned)
References
Details
(Whiteboard: [fidefe-sidebar])
Attachments
(2 files)
Found in
- Nightly 142.0a1 (2025-06-27)
Affected versions
- Nightly 142.0a1 (2025-06-27)
Affected platforms
- Ubuntu
Steps to reproduce
- Have at least 6 Pinned tabs. (More tabs easier to reproduce)
- Grab random pinned tabs and change their position by trying to move them to the end or the start of the Pinned section really fast.
- Grab the tabs from the edge of the pinned section and throw (I say throw because I lift my finger from the left mouse click before the tab lands into a new position) them to the middle. (Needs a mouse)
Expected result
- Pinned items should simply change position without issues.
Actual result
- The Pinned items start to overlap eachother and the entire UI is broken.
Regression range
Ill try to get a regression range for this issue since it doesnt happen in Beta or on Windows and Mac
Updated•9 months ago
|
Updated•9 months ago
|
Updated•9 months ago
|
Comment 1•9 months ago
|
||
I wonder what the connection between this and bug 1955112 ends up being, but I expect there to be one. Rares, can you actually get a reliable regression window here? I am a bit worried that we don't fully understand the STR just yet and that it's somewhat random.
Are there any errors in the browser console when this happens?
| Reporter | ||
Comment 2•9 months ago
|
||
Its hard to reproduce this issue, I usually keep trying until it happens, I dont have any real steps and no there are no errors in Console, I reproduced the issue in our latest Nightly builds and I did not see any errors.
Comment 4•8 months ago
|
||
Hi Rares, are you seeing this issue in the latest Nightly?
Updated•8 months ago
|
| Reporter | ||
Comment 5•8 months ago
|
||
Yes this issue still occurs in our latest Nighly 143.0a1 (2025-07-24)
Comment 6•8 months ago
|
||
(In reply to Rares Doghi, Desktop QA from comment #5)
Yes this issue still occurs in our latest Nighly 143.0a1 (2025-07-24)
Is this now also happening on 142 beta or no?
| Reporter | ||
Comment 7•8 months ago
|
||
Yes I was able to reproduce the issue in our latest Beta 142.0b2
Comment 8•8 months ago
|
||
:stransky, would you mind taking a look? Thanks and sorry. 😔
Comment 9•8 months ago
|
||
If bug 1955112 is related we hit a Wayland D&D issue here (Bug 1622107). Before bug 1955112 if there's unfinished D&D operation (D&D ended outside of Firefox window or double-tap or so?) we canceled the D&D operation. That led to unfinished D&D on Firefox UI and blocked user interaction here as UI expected input from D&D (was triggered by Bug 1957434 as it adds more check to D&D finish perhaps?)
So out Wayland D&D workaround for unfinished D&D was changed to actually finish the hanging D&D and it looks like it leads to this one - we're finishing D&D in the middle. IMHO if the workaround produces this outcome I think it may be reproduced by users as well without it by just dropping D&D early.
So right now we know that we're starting another D&D while the latest one was not finished:
https://searchfox.org/mozilla-central/rev/91b099f5033ae989b54d62c719442e3ecc5a1212/widget/gtk/nsWindow.cpp#7511
which we're hitting here. I just need to know which event we should deliver to front end to finish the pending D&D correctly (looks like both cancel or finish has its drawbacks or were not implemented correctly).
Description
•