Created attachment 357607 [details]
Display of the Candidate Window

Hi,

I cannot reproduce this bug. What's your libchewing version?

libchewing-0.3.2-15.fc11.x86_64

And below is my setting:
Editing:
    Auto move cursor:		    Off
    Add phrases in fron:	    Off
    Easy symbol input:		    On
    Esc clean all buffer:	    Off
    Maximum Chinese characters:	    20
    Force lowercase in En mode:	    Off
    Sync between CapsLock and IM:   keyboard

Selecting:
    Plain Zhuyin mode:		    Off
    Candidate per page:		    10
    Choose phrases from backward:   Off
    Space to select:		    On

Keyboard:
    Keyboard Type:	    default
    Selection keys:	    1234567890
    Hsu's selection key:    1

Try the following steps to reproduce:

1. LANG=zh_TW.UTF-8 /usr/bin/ibus-daemon -r
2. Open a gnome terminal
3. Select the input method of "Chewing" by typing Ctrl-Space.
4. Disable "Plain Zhuyin mode"
5. Press "ej3".
6. Press <Space> 6 times

Sorry, still cannot reproduce it.
Would you mind deleting ~/.chewing and try again?

BTW, I've made a new version of ibus-chewing, mind trying it?

Regards,

The abnormal characters in the candidate window still appear even if the directory ~/.chewing is removed.

OK.
Tell me how to do and I'll try it.

How about other input methods? And what your location (echo $LANG) ?

I don't know the situation of other input methods as I just use "chewing".

LANG=zh_TW.UTF-8

You can try to reproduce the bug with the steps mentioned in the comment #4.

I did follow your steps but still cannot reproduce the bug.

I see ibus-pinyin is also installed in your system. How about adding pinyin by:
1. Right click on ibus status icon->preference
2. on input methods tab -> select input method -> chinese -> pinyin
3. click on add and close the dialog.
4. Restart ibus-daemon by right click on ibus status icon->restart

Then, do following test:
1. Ctrl-space to enable IBus
2. switch to pinyin and type gu (which is equivalent to ㄍㄨ3)
3. See if the candidate selection dialog is messy.

Regards,
Ding-Yi Chen

Perhaps it is a font display problem.

I found that the strange characters in the candidate window will change when some different font size or windows resolution (DPI) is chosen.

Below is my test:
Font_Size____DPI____Result
==========================
Sans 12    81~101    Fail
Sans 12    80        OK
Sans 12    102       OK
Sans 10    89        Fail
Sans 10    90~97     OK
Sans 10    98~122    Fail
Sans 10    123       OK

My original setting is Sans 12, 96 dots per inch.
Therefore, I think I have to add an extra step to reproduce the problem.

1. LANG=zh_TW.UTF-8 /usr/bin/ibus-daemon -r
2. Open a gnome terminal
3. Select the input method of "Chewing" by typing Ctrl-Space.
4. Disable "Plain Zhuyin mode"
5. Press "ej3".
6. Press <Space> 6 times 
7. Change the Application font to "Sans 10" and 96 DPI by
   System -> Preferences -> Appearance -> Fonts
   Application font: Sans 10
   Details -> Resolution: 96 dots per inch

Hopefully, this problem can be reproduced.

Sorry, the last step is wrong.
The font size should be set to "Sans 12", not "Sans 10".

Following is the corrected steps:

7. Change the Application font to "Sans 12" and 96 DPI by
   System -> Preferences -> Appearance -> Fonts
   Application font: Sans 12
   Details -> Resolution: 96 dots per inch

(In reply to comment #9)
> I did follow your steps but still cannot reproduce the bug.
> 
> I see ibus-pinyin is also installed in your system. How about adding pinyin by:
> 1. Right click on ibus status icon->preference
> 2. on input methods tab -> select input method -> chinese -> pinyin
> 3. click on add and close the dialog.
> 4. Restart ibus-daemon by right click on ibus status icon->restart
> 
> Then, do following test:
> 1. Ctrl-space to enable IBus
> 2. switch to pinyin and type gu (which is equivalent to ㄍㄨ3)
> 3. See if the candidate selection dialog is messy.
> 
> Regards,
> Ding-Yi Chen  

Looks like the candidate selection dialog become messy when it contains the following characters:
鶻鼜淿鼔皷糓餶蛊䶜

Do you know how to enter those characters with pinyin?

Sorry, still cannot reproduce.
After removed wqy fonts, one of the candidate become a unicode number square, but otherwise is normal.

Can you use gucharmap to get the fonts for these character?

Usage:
1. run gucharmap &
2. search for one of those character.
3. Right click on the character. There will be a dialog pop up, and font information should be at the bottom.

Regards,
Ding-Yi Chen

I did run gucharmap as you said and found that the following character did screw up gucharmap:

糓
UTF-16: 0x7CD3

However, the warning message by gucharmap is as follows:
(gucharmap:2462): Pango-WARNING **: shaping failure, expect ugly output. shape-engine='BasicEngineFc', font='AR PL UMing HK Light 12', text='糔'

In addition, the same character (UTF-16:0x7CD3) also screw up the gnome-terminal if it is entered.

Just hope that someone can reproduce these problems based on the following conditions:
(1) platform: x86_64
(2) language: zh_TW.UTF-8 (Traditional Chinese)
(3) System -> Preferences -> Appearance -> Fonts
    Application font: Sans 12
    Details -> Resolution: 96 dots per inch

Still cannot reproduce. :-(

Maybe it is font rendering problem. Did you do any fontconfig customization? 

How about reinstall fontconfig?

I think it should be the font rendering problem.

I have tried to let more applications show that character (UTF-16:0x7CD3).
All will become messy when a specific font size is chosen.

I did "sudo yum reinstall fontconfig" but it didn't help.

I'll do a clean install of x86_64 FC11 with the language of zh_TW.UTF-8 and apply "yum update" to see if I can reproduce this problem.

By the way, back to the original input sequence "ej3" (ㄍㄨ3）,why the number of candidates varies with the selection of "Plain Zhuyin mode"?

The number of candidates is 20 when "Plain Zhuyin mode" is enabled.
And the number will be 59 if the "Plain Zhuyin mode" is disabled.

Is it what you expect?

(In reply to comment #17)
> By the way, back to the original input sequence "ej3" (ㄍㄨ3）,why the number of
> candidates varies with the selection of "Plain Zhuyin mode"?
> 
> The number of candidates is 20 when "Plain Zhuyin mode" is enabled.
> And the number will be 59 if the "Plain Zhuyin mode" is disabled.
> 
> Is it what you expect?  

It is a bug which I can reproduce. :-)
Do you mind if I change the component to fontconfig for font rendering bug,
while file a new bug for the plain Zhuyin bug?

I have done two clean installs of Fedora 11, x86_64 and i386 respectively.  Both are using "Traditional Chinese" during the install process and apply the latest update after installation.

I take a look at the font size and DPI setting on both of them and they are the same as listed below:
應用程式字型：Sans,10
文件字型：    Sans,10
桌面字型：    Sans,14
視窗標題字型：Sans Bold 10
固定寬度字型：Monospace 10
解析度：96 DPI

Then I open "gucharmap" and search the character "糓" (U+7CD4).
Change the font size to 12, 11 and 10.

The i386 version works quite well.

However, the display on the x86_64 one become messy.
And the following warning message appear on the terminal:
(gucharmap:1997): Pango-WARNING **: shaping failure, expect ugly output. shape-engine='BasicEngineFc', font='AR PL UMing HK Light 12', text='糔'

(gucharmap:1997): Pango-WARNING **: shaping failure, expect ugly output. shape-engine='BasicEngineFc', font='AR PL UMing HK Light 11', text='糔'

(gucharmap:1997): Pango-WARNING **: shaping failure, expect ugly output. shape-engine='BasicEngineFc', font='AR PL UMing HK Light 9.9990234375', text='糔'

(In reply to comment #18)
> (In reply to comment #17)
> > By the way, back to the original input sequence "ej3" (ㄍㄨ3）,why the number of
> > candidates varies with the selection of "Plain Zhuyin mode"?
> > 
> > The number of candidates is 20 when "Plain Zhuyin mode" is enabled.
> > And the number will be 59 if the "Plain Zhuyin mode" is disabled.
> > 
> > Is it what you expect?  
> 
> It is a bug which I can reproduce. :-)
> Do you mind if I change the component to fontconfig for font rendering bug,
> while file a new bug for the plain Zhuyin bug?  

I prefer filing a new font-rendering bug.
Mark this one as its duplicate.
The new one will contain less but clearer information.

Another issue about "plain Zhuyin bug" is in the case that the "Plain Zhuyin mode" is enabled and the desired candidate is not in the first window.

For example, when the candidate in the first window is selected, it will go directly to the output.
If the candidate not in the first window is selected, it will stay in the buffer, not going out to the output.

(In reply to comment #20)
> (In reply to comment #18)
> > (In reply to comment #17)
> > > By the way, back to the original input sequence "ej3" (ㄍㄨ3）,why the number of
> > > candidates varies with the selection of "Plain Zhuyin mode"?
> > > 
> > > The number of candidates is 20 when "Plain Zhuyin mode" is enabled.
> > > And the number will be 59 if the "Plain Zhuyin mode" is disabled.
> > > 
> > > Is it what you expect?  
> > 
> > It is a bug which I can reproduce. :-)
> > Do you mind if I change the component to fontconfig for font rendering bug,
> > while file a new bug for the plain Zhuyin bug?  
> 
> I prefer filing a new font-rendering bug.
> Mark this one as its duplicate.
> The new one will contain less but clearer information.

Please review Bug 519317 and comment on it.

> Another issue about "plain Zhuyin bug" is in the case that the "Plain Zhuyin
> mode" is enabled and the desired candidate is not in the first window.
> 
> For example, when the candidate in the first window is selected, it will go
> directly to the output.
> If the candidate not in the first window is selected, it will stay in the
> buffer, not going out to the output.  

Plain Zhuyin bug is now at Bug 519328.

*** This bug has been marked as a duplicate of bug 519317 ***