1541921 - Quantum Bar: When typing a letter and deleting it immediately again makes the visited pages quick access Drop Down appear

Status: VERIFIED FIXED

(Firefox :: Address Bar, defect, P3)

Description

[Mehmet]

Attachment: screencast.mov

Firefox Nightly 68.0a1 (2019-04-04) (64-Bit)
MacOS 10.14.4 - but probably OS=ALL_Desktop

STR:

1.) Enable Quantum Bar or be in the Experiment
2.) Visit some pages so that the arrow down button (visited pages quick access button?) includes links
3.) Now focus the Address Bar and type one letter and immediately delete this letter again

Actual: The arrow down button (visited pages quick access button?) is activated and the Drop Down with the visited links appears

Expected: It should not appear. It should only appear when you click on the Arrow Down button

A screencast is attached.

Thanks.

Comment 1

[Marco Bonardo [:mak]]

it looks like a race condition the popup opens for the "a" search but it actually searches for "".

Comment 2

[Drew Willcoxon :adw]

Attachment: Bug 1541921 - Remove duplicate-search bailout from quantumbar.

We need to reset _lastSearchString in _on_input when the user backspaces over the last char. There are a couple of problems in this bug. The first is that the popup doesn't open and a new search doesn't start when you type the same char for the second time, after having backspaced over it. That's because you hit the same-search early return in _on_input because _lastSearchString hasn't been updated by that point. It's still the old value, the single char. The second problem follows from the first, and it's that when you then backspace over the char, the history popup opens because view.isOpen is false, so the early return in _on_input that checks for an empty search string (and open view) is not hit, and we continue on to do an empty search.

Comment 3

[Drew Willcoxon :adw]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=9a9729796062dd74e95c5268dc44db67d3af357b

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/b97df75c9365
Remove duplicate-search bailout from quantumbar. r=dao

Comment 5

[Cristian Brindusan [:cbrindusan]]

Backed out changeset b97df75c9365 (Bug 1541921) for bc mochitest asan failures.

Backout: https://hg.mozilla.org/integration/autoland/rev/2b51f2ae2a5c53a46050cda19561b5f504348cb9

Push that started the failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=pending%2Crunning%2Csuccess%2Ctestfailed%2Cbusted%2Cexception&revision=b97df75c9365a162675d962354083af7b5fb40b9&selectedJob=241510882

Failure log: https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=241510882&repo=autoland&lineNumber=4057

Comment 6

[Drew Willcoxon :adw]

I have no idea how this simple patch causes an ASAN failure. It didn't happen on try, but I rebased on the current tree before landing. Another try push:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=de1d079ff28d23187dbeea87d5edec723289ab7d

Comment 7

[Drew Willcoxon :adw]

Oh, maybe it's not actually an ASAN problem, and instead Firefox just wouldn't shut down, so this terminated it (with an intentional segfault?): https://searchfox.org/mozilla-central/source/toolkit/components/terminator/nsTerminator.cpp

Comment 8

[Drew Willcoxon :adw]

Failing on try now too. Maybe it's the new test, although it must be in conjunction with something else since I last updated my tree. And I doubt it since it's not in the failing logs. But anyway, without it: https://treeherder.mozilla.org/#/jobs?repo=try&revision=6519d6ed172d8b49ac7f65cbf956d1d93b13ff6d

Comment 9

[Drew Willcoxon :adw]

I haven't been able to figure this out today. I'm starting to think something is wrong with bug 1522278's fix, especially after bug 1545731 was filed, and this bug somehow exposed it. But I'm not sure. Here's a try push reverting that bug and with the patch to this bug applied: https://treeherder.mozilla.org/#/jobs?repo=try&revision=c00a452492fda185835e6aea46014b801fcfc819

Comment 10

[Drew Willcoxon :adw]

I tried bisecting mozilla-central with the patch applied and pushing to try to see which revision caused the difference between before I refreshed my tree and after. Long story short, other failures introduced in various revisions and subsequent backouts made that hard to do, but I found a revision, but it doesn't make much sense because it has to do with Servo layout. So I'm not very confident about it.

As a sanity check, I refreshed my tree tonight and pushed the patch to try again, and there's no problem: https://treeherder.mozilla.org/#/jobs?repo=try&revision=ebf062666325f31b60330ba37d8ab94a98ab6c56

So I'm just going to land this again.

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/mozilla-central/rev/cd4bab29d164
Remove duplicate-search bailout from quantumbar. r=dao

Comment 12

[Bogdan Tara[:bogdan_tara | bogdant]]

https://hg.mozilla.org/mozilla-central/rev/cd4bab29d164

Comment 13

[Alexandru Trif, Desktop QA [:atrif]]

Hello,
Reproduced the issue using Firefox 68.0a1 (20190404215521) on Windows 10 x64 and macOs 10.14.
The issue is verified fixed using Firefox 68.0b13 (20190624133534) and Firefox 69.0a1 on Windows 10x64, macOS 10.14, Ubuntu 18.04.