1545731 - Don't autofill on backspace when a non-heuristic result is selected

Status: RESOLVED DUPLICATE of bug 719888

(Firefox :: Address Bar, defect, P3)

Description

[Dão Gottwald [:dao]]

+++ This bug was initially created as a clone of Bug #1522278 +++

See bug 1522440 comment 20:

With bug 1543561 fixed the test now times out here: https://searchfox.org/mozilla-central/rev/d302c3058330a57f238be4062fddea629311ce66/accessible/tests/browser/events/browser_test_focus_urlbar.js#155

The reason seems to be that we incorrectly autofill after backspace, and then the result list is too short, and instead of selecting a result the test selects a one-off search button.

It looks like bug 1522278 wasn't fully effective if it didn't even cause this bug in the first place.

Comment 1

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

Bugbug thinks this bug is a task, but please change it back in case of error.

Comment 2

[Mike de Boer [:mikedeboer]]

Bot is wrong.

Comment 3

[Drew Willcoxon :adw]

Even with bug 1522278 reverted, this test still causes autofill on backspace. (I noticed that the test also produces different results with quantumbar vs. awesomebar even up to the point where the autofill incorrectly happens, though, so I'm not clear on how much of the test's hanging on quantumbar has to do with that vs. this autofill problem.)

The test presses the down arrow key to select a non-heuristic result and then backspaces, and then autofill is triggered.

The problem before bug 1522278 landed (I think) is that the autofill logic only did prefix matching against the autofilled placeholder. In the case of the test, the placeholder is irrelevant.

After bug 1522278 landed, there's no selection in the input at all, subverting the new don't-autofill logic.

Comment 4

[Drew Willcoxon :adw]

Removing bug 1522278 in the regressed-by field, see previous comment.

Comment 5

[Drew Willcoxon :adw]

I just noticed awesomebar has this same behavior -- autofilling on backspace after selecting a non-heuristic result -- so this isn't even something that quantumbar regressed. (Even so, we should still consider fixing it.) So it seems the reason this test is now hanging is that different results are produced on quantumbar vs. awesomebar up to the point that the test hangs on quantumbar.

Comment 6

[Drew Willcoxon :adw]

Also removing this as a quantumbar-release blocker since this happens in awesomebar, as mentioned earlier. Actually we should either close this bug or morph it into not autofilling on backspace when a non-heuristic result is selected.

Comment 7

[Drew Willcoxon :adw]

This isn't urgent, so I'm going to unassign myself and change the priority.

Comment 9

[Drew Willcoxon :adw]

This is bug 719888 actually