24464 - IME-Japanese chars will be displayed the wrong place

Status: VERIFIED FIXED

(Core :: Internationalization, defect, P3)

Description

[Teruko Kobayashi]

After you type Japanese characters in the text field and delete the characters,
Japanese characters you type again will start showing up at the point the
previous Japanese characters are deleted.

Steps of reproduce
1. Go to the above URL
2. Turn on IME (Kinput2) in linux
3. In Name field after Customer ID field, type 3 Japanese characters
4. Use Backspace key (hit backspace key 3 times) to delete these 3 Japanese
characters
5. Type Japanese characters again
Notice that the uncommitted Japanese characters will be displayed the place
the previous Japanese characters were deleted.  After you commit the characters,
the Japanese characters are displayed at the beginning of the text field.

Tested 2000011908 Linux build.
5.

Comment 1

[Frank Tang]

Change platform to PC

Comment 2

[Frank Tang]

I believe this is still a XIC focus issue.

Comment 3

[Masaki Katakai]

During we (I and Toshi) were working on new XIM codes (on-the-spot style),
we found a problem that timer for updating spot location is stopped when
mCursorPosition.x = mCursorPosition.y = 0.

This happens when the cursor is located at beginning of text field.
Once the cursor is moved to the beginning of text field, the timer for
the text field will be stopped, and will not be started until next focus-in.

*** nsWidget.cpp.new    Mon Feb 14 13:39:11 2000
--- nsWidget.cpp        Thu Feb  3 14:51:10 2000
***************
*** 122,129 ****
     compEvent.theReply.mCursorPosition.y=-1;
     DispatchEvent(&compEvent, status);
     // set SpotLocation
!    if((compEvent.theReply.mCursorPosition.x < 0) &&
!       (compEvent.theReply.mCursorPosition.y < 0))
       return NS_ERROR_FAILURE;
     if((compEvent.theReply.mCursorPosition.x != oldx)||
        (compEvent.theReply.mCursorPosition.y != oldy))
--- 122,129 ----
     compEvent.theReply.mCursorPosition.y=-1;
     DispatchEvent(&compEvent, status);
     // set SpotLocation
!    if((compEvent.theReply.mCursorPosition.x <= 0) &&
!       (compEvent.theReply.mCursorPosition.y <= 0))
       return NS_ERROR_FAILURE;
     if((compEvent.theReply.mCursorPosition.x != oldx)||
        (compEvent.theReply.mCursorPosition.y != oldy))

I was not able to access the URL in bug report, so I could not verify
our fix on the URL. But I believe our changes will fix this problem.

Comment 4

[Teruko Kobayashi]

Attachment: test case

Comment 5

[Hidetoshi Tajima]

I tested the attachedment 5238 with the cvs build. Without
Katakai's fix above, I could reproduce the problem on any
text field there, but with the fix compiled in, then the
problem doesn't occur.

Comment 6

[Frank Tang]

fix and check in.

Comment 7

[Teruko Kobayashi]

Timing issue is filed in bug 28029. I verified this in 2000021608 Linux build.