Acroread 8.x CJK input problem

Question

See also https://bugzilla.novell.com/show_bug.cgi?id=353251
CJK input doesnt work right in the English version of
acroead 8.1.1 or the pre-release version 8.1.2.
First of all, one needs to set
export GTK_IM_MODULE=xim
to make acroread react at all to the hotkey which triggers
SCIM input (default hotkeys on openSUSE are Shift+Space and Control+Space).
When GTK_IM_MODULE=scim or GTK_IM_MODULE=scim-bridge, acroread
wont react to the hotkey which enables SCIM at all.
With GTK_IM_MODULE=xim, the scim input method *can* be
enabled by Shift+Space. But it doesnt work right.
One can see correct Japanese in the popups shown by scim when
converting phonetics to Chinese characters. But the predit string
shows garbage which seems to resemble Arabic. And after comitting
everything typed is converted to question marks (acroread 8.1.1)
or boxes (acroread 8.1.2).
I am talking about the *English* versions of acroread here, not
localized Japanese versions like e.g.
ftp://ftp.adobe.com/pub/adobe/reader/unix/8.x/8.1.1/jpn
As it is not nice to have completely packages for each language, I
hope that the different language versions can be merged into one in
the long run.
It would be very nice if there were only one basic version and to
support other languages one only had to add fonts and translations and
not exchange the binaries.

Answer 1

Some more comments from: https://bugzilla.novell.com/show_bug.cgi?id=353251
------- Comment #5 From Gaurav Jain 2008-01-11 10:48:18 MST -------
[ ] Private
Mike,
Could you confirm if everything is working fine with the Japanese Reader
downloaded from the Reader website? That'll be really strange since the Viewer
binaries should be identical in the English and Japanese versions. The only
difference in the 2 installers should be in the fonts and the resource
libraries.
-vc
------- Comment #6 From Mike Fabian 2008-01-11 19:43:16 MST -------
[ ] Private
OK, I tried with the special Japanese version of acroread (8.1.1)
as well.
There is no difference in behaviour between the Japanese acroread
8.1.1 and the English acroread 8.1.1 and the English acroread 8.1.1 as
far as the input problem reported here is concerned! All of them show
the problem as reported here.
But I found that it depends on the locale. If acroead is started
in ja_JP.UTF-8 locale
LANG=ja_JP.UTF-8 acroread
the problem occurs as reported here.
However, if acroread is started in ja_JP.eucJP locale
LANG=ja_JP.eucJP acroread
the Japanese input works fine! Thats the same with the Japanese and
the English version of acroread.
As UTF-8 locales are the default nowadays on most Linux distributions,
it is important that this works not only in legacy locales like
ja_JP.eucJP but also in ja_JP.UTF-8.
------- Comment #7 From Gaurav Jain 2008-01-13 09:25:27 MST -------
[ ] Private
The screen-shot seems to indicate you are trying to enter japanese characters
in a standard GTK+ edit field. The core Reader code actually doesn't interact
much with the control during the process of entry of text.
Could you try the following -
1. When the default locale is eucJP, what happens in a standard edit field in
some other GTK+ app., for instance gtk-demo? If you don't have gtk-demo, you
could even try the same thing in the Open dialog in the Adobe Reader, where you
type the file name.
2. If you don't export GTK_IM_MODULE=xim, can you make the IME appear in some
other GTK+ app. like the gtk-demo?
My guess is this (atleast point 1 above) may be a problem with GTK+, though we
are investigating this at our end as well. Maybe an issue with the fonts that
get loaded when the locale is UTF-8, vs eucJP. That's the reason, the
characters loaded from the IME are showing as question marks.
-vc
------- Comment #8 From Mike Fabian 2008-01-14 05:18:46 MST -------
[ ] Private
Gaurav Jain> 1. When the default locale is eucJP, what happens in a
Gaurav Jain> standard edit field in some other GTK+ app., for instance
Gaurav Jain> gtk-demo?
In gtk-demo, Japanese input works both for ja_JP.eucJP locale
*and* for ja_JP.UTF-8 locale.
And it works for all values of GTK_IM_MODULE which I tried
(GTK_IM_MODULE=xim, GTK_IM_MODULE=scim, and GTK_IM_MODULE=scim-bridge.
Gaurav Jain> If you don't have gtk-demo, you could even try
Gaurav Jain> the same thing in the Open dialog in the Adobe Reader,
Gaurav Jain> where you type the file name.
Japanese input in the Open dialog of the Adobe Reader behaves
exactly like in the search field of the Adobe Reader:
- works fine in ja_JP.eucJP locale
(for all values of GTK_IM_MODULE)
- does not work in ja_JP.UTF-8 locale
(not for any of the above mentioned values of GTK_IM_MODULE)
Gaurav Jain> 2. If you don't export GTK_IM_MODULE=xim, can you make
Gaurav Jain> the IME appear in some other GTK+ app. like the gtk-demo?
In gtk-demo the IME appears and works fine.
In acroread, the IME appears for all values of GTK_IM_MODULE (xim,
scim, scim-bridge) and input works fine in ja_JP.eucJP locale. In
ja_JP.UTF-8 locale, the IME appears as well and input seems to be
possible but the result is garbage as in my screen shot.
------- Comment #9 From Gaurav Jain 2008-01-22 04:06:32 MST -------
[ ] Private
Hello Mike,
I tried reproducing the bug on SLED 10, openSUSE 10.3, and openSUSE
11.0(http://download.opensuse.org/distribution/11.0-Alpha1/iso/cd/openSUSE-11.0-Alpha1-GNO ME-i386.iso)
but was unable to do so.I could see no difference in the behavior based on
locale.Also,the problem of garbage predit strings in case of utf8 is not
reproducible at my end.
Could you please attach a screenshot of your SCIM setup settings,and the
download location of openSUSE 11.0.
Regards,
Sanika
------- Comment #10 From Mike Fabian 2008-01-23 10:48:40 MST -------
[ ] Private
I found that the problem occurs *only* with GTK_IM_MODULE=xim,
contrary to what I wrote in comment #8.
That was my fault, because I still had
# Workaround for http://rudin.suse.de:8888/show_bug.cgi?id=85416
# (see comment #37):
export GTK_IM_MODULE=xim
patched into the beginn of the acroread start-script.
Apparently this is not needed anymore, input using
GTK_IM_MODULE=scim and GTK_IM_MODULE=scim-bridge seems to
work fine now in acroread 8.1.2.
*But* the problem I described here occurs with GTK_IM_MODULE=xim.
Sanika,
can you reproduce the problem with GTK_IM_MODULE=xim ?