447945 - IME candidate list is positioned to wrong place when the caret is in <panel>

Status: RESOLVED FIXED

(Core :: Widget: Gtk, defect)

Description

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

spinning off from bug 433340.

On Linux, the candidate list window of IME is positioned to wrong place. On GTK2, Gecko returns the caret position that is the rendering view relative coordinates. In <panel>, the rendering view is for the <panel> element. However, the IME events are dispatched by the parent window. Therefore, we fail to convert the caret position.

Comment 1

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Attachment: Patch v2.0

This is simplest approach.

IME events and text event should return the widget that is the base of the caret position.

Then, widget can recompute the caret position from the actual focused widget to IME context owner widget.

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

Canceling the review. Because I have a report for this patch in bugzilla-jp. I need to check the reported issue.

Comment 3

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

sorry for the spam. this patch doesn't have problem, re-requesting the reviews.

Comment 4

[Masaki Katakai]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

looks OK for me.

Comment 5

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

pushed to trunk.

if we don't have any regression reports, I'll request a1.9.0.x.

Comment 6

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

I don't have any regression reports of this patch. Let's take this patch to 1.9.0 branch.

This bug is very unusable for IME users of Linux.

Comment 7

[Samuel Sidler (old account; do not CC)]

Can we get a test for this bug? I know cocoa widgets can't, but hopefully GTK can be tested...

Comment 8

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

(In reply to comment #7)
> Can we get a test for this bug? I know cocoa widgets can't, but hopefully GTK
> can be tested...

I have no idea. However, this behavior is only in Gecko, so, we can make pseudo testing without IME, maybe.

Comment 9

[Samuel Sidler (old account; do not CC)]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

Moving this out to 1.9.0.3. We really need a test here...

Comment 10

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Samuel:

I'm still thinking the way for creating the automated test. However, maybe, we need to create new APIs like key hell tests...

Comment 11

[Samuel Sidler (old account; do not CC)]

Masayuki: If you need new APIs, please file bugs on those and CC me. Thanks. :)

Comment 12

[Samuel Sidler (old account; do not CC)]

Masayuki: Any update here?

Comment 13

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

(In reply to comment #12)
> Masayuki: Any update here?

I'm still think for the way to test. But I'll file the bug soon.

Comment 14

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(away: 6/25-6/28)]

Samuel:

I filed a meta bug for IME automated tests, it's bug 460056. But I don't have actual idea for implement this testing...

Comment 15

[Daniel Veditz [:dveditz]]

Comment on attachment 331403 [details] [diff] [review]
Patch v2.0

This sounds cosmetic rather than functional, and not even any screenshots for QA to verify against. Not approving for the 1.9.0 branch, this doesn't appear to meet the branch criteria and 3.1 is coming soon.