226784 - Caret display problems with Korean input methods in Windows

Status: VERIFIED FIXED

(Core :: Widget: Win32, defect)

Description

[SooBong]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 Firebird/0.7
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 Firebird/0.7


Problem : Caret is displayed to the left side of Korean characters during
pre-edit. This is bothering Korean users because the caret is usually either not
visible in pre-edit or placed to the right side of the pre-edit area.

Cause of the problem : Windows IME. Windows Korean IME returns the caret
position to the left side of the korean but it has been OK since the pre-edit
area is always inverted, so it actually hides the caret or some applications
disables the caret while composition is going on.

Suggestion : We can not change the behavior of Korean IME in Windows, so hiding
the caret while composition is going on, and display it after committed. 


Reproducible: Always

Steps to Reproduce:
1. Install Firebird 0.7 to windows 2000
2. Type in some korean character into a textfield
3. Observe the cursor displaying to the left of the pre-edit



Expected Results:  
Suggestion : We can not change the behavior of Korean IME in Windows, so hiding
the caret while composition is going on, and displaying it after committed.

Comment 1

[Isaac Hwak Han]

Confirming with Gecko/20040119 on Korean WinXP using MS IME 2003.

Severity to minor, because there is no loss of function from this bug.

Comment 2

[Martijn Wargers (dead)]

Is this still an issue with current trunk build?
ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk

Comment 3

[Jaesik]

Yes, the issue is still alive on firefox 3.0.4. Nobody working on it? It has no loss of function but quite uncompotable to input Korean characters. On Linux, this issue does not happen.

Comment 4

[Jaesik]

Yes, the issue is still alive on firefox 3.0.4. Nobody working on it? It has no
loss of function but quite uncomfotable to input Korean characters. On Linux,
this issue does not happen.

Comment 5

[Martijn Wargers (dead)]

Could you perhaps test with a trunk build?
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-mozilla-central/

Comment 6

[Jaesik]

Trunk build still keeps the issue.

Comment 7

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

The Korean IME which is default IME on Vista returns 0 for cursor position via ImmGetCompositionStringW. And also clause length is also 0. The IME's behavior is bad :-(

Comment 8

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Patch v1.0

Korean IME doesn't return clause information correctly. Then, we are processing the composition string specially. Then, the cursor pos should be end of the composition string, probably.

Comment 9

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

If this is risky, we can add the limitation for the new code:

* the current keyboard layout is 949.
* the current composition string length is 1.
* the current cursor position is 0.

Do we have other conditions?

Comment 10

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Patch v2.0

Thank you, Kimura-san for your review in bugzilla-jp.

I think that we need to change IME caret undefined case at (sIMECompClauseLength > 0) case too. And also I add notice comments to nsGUIEvent.h and nsIPrivateTextRange.h.

Comment 11

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Patch v2.1

oops, we should check the return value of ::ImmGetCompositionString.

Comment 12

[Masatoshi Kimura [:emk]]

Comment on attachment 352403 [details] [diff] [review]
Patch v2.1

> I think that we need to change IME caret undefined case at
> (sIMECompClauseLength > 0) case too.
Good catch.

Comment 13

[Ere Maijala (slow)]

Comment on attachment 352403 [details] [diff] [review]
Patch v2.1

+#define NS_TEXTRANGE_RAWINPUT              0X02

Change the X to lowercase while at it.

Comment 14

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Patch v2.1.1

fix comment 13, sorry for the spam.

Comment 15

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 352974 [details] [diff] [review]
Patch v2.1.1

+  // Note that the range array may not have an express caret position.
+  // So, an array may not have TEXTRANGE_CARETPOSITION typed range.

  // Note that the range array may not specify a caret position; in that
  // case there will be no range of type TEXTRANGE_CARETPOSITION in the array.

Comment 16

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: Patch v2.1.2

Comment 17

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

checked-in to trunk
http://hg.mozilla.org/mozilla-central/rev/79136a417a3a

Comment 18

[Martijn Wargers (dead)]

Isaac, could you perhaps test in current trunk builds, to see if this is now fixed for you? ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk
Masayuki, thanks for fixing this!

Comment 19

[Jaesik]

It is fixed. Much better to use. Thanks Masayuki.

Comment 20

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Thank you for your verification.

Comment 21

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Comment on attachment 353051 [details] [diff] [review]
Patch v2.1.2

This patch only changes the behavior of some IMEs which doesn't support caret position property. Most IMEs support it. I don't know such IMEs other than the standard Korean IME.

Comment 22

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 353051 [details] [diff] [review]
Patch v2.1.2

a191=beltzner

Comment 23

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

pushed to 1.9.1
http://hg.mozilla.org/releases/mozilla-1.9.1/rev/5dec81af74a7

Comment 24

[Tony Chung [:tchung]]

Per comment #19, marking verified for trunk.

Can someone with korean IME verify this on the 1.9.1 branch?  Builds here: ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-mozilla-1.9.1-l10n/

Comment 25

[Hidehiro Kozawa]

(In reply to comment #24)
> ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-mozilla-1.9.1-l10n/
verified 1.9.1