Re: Rendering Fix progresses, Now Input-Method

Re: Rendering Fix progresses, Now Input-Method

[Shyam | ശ്യാം കാരനാട്ട് | Karanattu]

Kenichi Handa <handa@m17n.org> writes:

> Are those keys already widely used by Malayalam people?
> If so, we'll add them in quail/indian.el.
Ofcourse, it is widely used in all indian languages. For example അവന്
means "for him" അവന്‍  means "he" the difference is just that zwj and the
rendering will also change to show another character. It is similarly
used for all indian language and actually should be added to
inscript-dev-keytable itself. and its indispensable for the keymap
support of any indian language.
> Ah, I've just installed a fix.

Thanks!
> Please show me the output of gdb's "bt" command at that
> moment.  And I think the problem is not in the input method
> but in the display routine of the resulting character
> sequence.  Please tell me which font is used and which
> character sequence (preferably the shortest one) causes this
> bug so that I can reproduce it.
>
These sequence causing crash are not any usage of
malayalam(but there may be), they are
just a random key presses of the choice of keystrokes available;-)
Below is the result of what i get as i open emacs and do the
following
C-u C-\ malayalam-inscript RET qea
where q is ൌ
e is ാ
a is ൊ

Starting program: /home/shyam/emacs/src/emacs 
[Thread debugging using libthread_db enabled]
[New Thread 0xb7143720 (LWP 10288)]

Program received signal SIGABRT, Aborted.
[Switching to Thread 0xb7143720 (LWP 10288)]
0xb7fde410 in __kernel_vsyscall ()
(gdb) bt
#0  0xb7fde410 in __kernel_vsyscall ()
#1  0xb757d4b6 in kill () from /lib/tls/i686/cmov/libc.so.6
#2  0x08117f8b in abort () at emacs.c:424
#3  0x08141295 in buf_charpos_to_bytepos (b=0x8383218, charpos=10288)
    at marker.c:140
#4  0x081d1706 in composition_update_it (cmp_it=0xbfcfb2dc, charpos=196, 
    bytepos=198, string=137874713) at composite.c:1198
#5  0x0807d3ae in next_element_from_composition (it=0xbfcfaf44)
    at xdisp.c:6611
#6  0x0807d735 in next_element_from_buffer (it=0xbfcfaf44) at xdisp.c:6496
#7  0x0807aa4f in get_next_display_element (it=0xbfcfaf44) at xdisp.c:5655
#8  0x0808091e in display_line (it=0xbfcfaf44) at xdisp.c:16578
#9  0x0808a3b2 in try_window (window=141387852, pos=
      {charpos = 1, bytepos = 1}, check_margins=1) at xdisp.c:14042
#10 0x0808befc in redisplay_window (window=141387852, just_this_one_p=0)
    at xdisp.c:13665
#11 0x0808d603 in redisplay_window_0 (window=141387852) at xdisp.c:12255
#12 0x081819d6 in internal_condition_case_1 (
    bfun=0x808d5e0 <redisplay_window_0>, arg=141387852, handlers=137861477, 
    hfun=0x8068a40 <redisplay_window_error>) at eval.c:1559
#13 0x08078b96 in redisplay_windows (window=6) at xdisp.c:12234
#14 0x0808e0ae in redisplay_internal (
    preserve_echo_area=<value optimized out>) at xdisp.c:11810
#15 0x081251b6 in read_char (commandflag=1, nmaps=2, maps=0xbfcfcf90, 
    prev_event=137874713, used_mouse_menu=0xbfcfd020, end_time=0x0)
    at keyboard.c:2649
#16 0x0812729f in read_key_sequence (keybuf=0xbfcfd0c4, bufsize=30, 
    prompt=137874713, dont_downcase_last=0, can_return_switch_frame=1, 
    fix_current_buffer=1) at keyboard.c:9343
#17 0x08129267 in command_loop_1 () at keyboard.c:1621
#18 0x08181c00 in internal_condition_case (bfun=0x81290c0 <command_loop_1>, 
    handlers=137917921, hfun=0x8124000 <cmd_error>) at eval.c:1511
#19 0x08123553 in command_loop_2 () at keyboard.c:1338
#20 0x08181cda in internal_catch (tag=137913897, 
    func=0x8123530 <command_loop_2>, arg=137874713) at eval.c:1247
#21 0x08123e67 in command_loop () at keyboard.c:1317
#22 0x081241db in recursive_edit_1 () at keyboard.c:942
#23 0x08124311 in Frecursive_edit () at keyboard.c:1004
#24 0x08118d8c in main (argc=1, argv=0xbfcfd744) at emacs.c:1728
(gdb) 
The above 
similarly i get many hangs for other key combinations like
C-u C-\ malayalam-inscript RET kra
there k can be many other letter probably those ones which can join with
"a"i.e. a="ോ" for me.
interestingly this crash dont happen with emacs' own keymap. But only
with the xkb's layout.
k is ക but that can also be i: ഗ but not D:അ 
t is ൂ
anyway all time, "a"(ോ) is creating some problems like if ae or qw makes the characters maped
to e and w to display twice. May be a is giving an extra space
considering the letter to be included usually between the two parts of
"ോ"
i am using meera and surprisingly with rachana font, the above specific
bugs are not there.. still it too crashes at some point which i am yet
to track with..
Other than crashes, the display also gets weird at some time for some
random key sequence with malayalam characters
"രോപ്ുോപര്ുേതോപ്തകോര" this is an example of such a thing.
The picture below shows the rendering difference between that in emacs
and in gedit.
   But emacs too makes the rendering perfect at times like when that
   text is selected for yanking. The gedit too is not perfect in rendering the
   above sequence. Avoiding those dotted circles are the thing avoided
   to make perfect rendering for gnome.


Thanks a lott
Shyam K

Re: Rendering Fix progresses, Now Input-Method

[Shyam | ശ്യാം കാരനാട്ട് | Karanattu]

Kenichi Handa <handa@m17n.org> writes:

> Are those keys already widely used by Malayalam people?
> If so, we'll add them in quail/indian.el.
Ofcourse, it is widely used in all indian languages. For example അവന്
means "for him" അവന്‍  means "he" the difference is just that zwj and the
rendering will also change to show another character. It is similarly
used for all indian language and actually should be added to
inscript-dev-keytable itself. and its indispensable for the keymap
support of any indian language.
> Ah, I've just installed a fix.

Thanks!
> Please show me the output of gdb's "bt" command at that
> moment.  And I think the problem is not in the input method
> but in the display routine of the resulting character
> sequence.  Please tell me which font is used and which
> character sequence (preferably the shortest one) causes this
> bug so that I can reproduce it.
>
These sequence causing crash are not any usage of
malayalam(but there may be), they are
just a random key presses of the choice of keystrokes available;-)
Below is the result of what i get as i open emacs and do the
following
C-u C-\ malayalam-inscript RET qea
where q is ൌ
e is ാ
a is ൊ

Starting program: /home/shyam/emacs/src/emacs 
[Thread debugging using libthread_db enabled]
[New Thread 0xb7143720 (LWP 10288)]

Program received signal SIGABRT, Aborted.
[Switching to Thread 0xb7143720 (LWP 10288)]
0xb7fde410 in __kernel_vsyscall ()
(gdb) bt
#0  0xb7fde410 in __kernel_vsyscall ()
#1  0xb757d4b6 in kill () from /lib/tls/i686/cmov/libc.so.6
#2  0x08117f8b in abort () at emacs.c:424
#3  0x08141295 in buf_charpos_to_bytepos (b=0x8383218, charpos=10288)
    at marker.c:140
#4  0x081d1706 in composition_update_it (cmp_it=0xbfcfb2dc, charpos=196, 
    bytepos=198, string=137874713) at composite.c:1198
#5  0x0807d3ae in next_element_from_composition (it=0xbfcfaf44)
    at xdisp.c:6611
#6  0x0807d735 in next_element_from_buffer (it=0xbfcfaf44) at xdisp.c:6496
#7  0x0807aa4f in get_next_display_element (it=0xbfcfaf44) at xdisp.c:5655
#8  0x0808091e in display_line (it=0xbfcfaf44) at xdisp.c:16578
#9  0x0808a3b2 in try_window (window=141387852, pos=
      {charpos = 1, bytepos = 1}, check_margins=1) at xdisp.c:14042
#10 0x0808befc in redisplay_window (window=141387852, just_this_one_p=0)
    at xdisp.c:13665
#11 0x0808d603 in redisplay_window_0 (window=141387852) at xdisp.c:12255
#12 0x081819d6 in internal_condition_case_1 (
    bfun=0x808d5e0 <redisplay_window_0>, arg=141387852, handlers=137861477, 
    hfun=0x8068a40 <redisplay_window_error>) at eval.c:1559
#13 0x08078b96 in redisplay_windows (window=6) at xdisp.c:12234
#14 0x0808e0ae in redisplay_internal (
    preserve_echo_area=<value optimized out>) at xdisp.c:11810
#15 0x081251b6 in read_char (commandflag=1, nmaps=2, maps=0xbfcfcf90, 
    prev_event=137874713, used_mouse_menu=0xbfcfd020, end_time=0x0)
    at keyboard.c:2649
#16 0x0812729f in read_key_sequence (keybuf=0xbfcfd0c4, bufsize=30, 
    prompt=137874713, dont_downcase_last=0, can_return_switch_frame=1, 
    fix_current_buffer=1) at keyboard.c:9343
#17 0x08129267 in command_loop_1 () at keyboard.c:1621
#18 0x08181c00 in internal_condition_case (bfun=0x81290c0 <command_loop_1>, 
    handlers=137917921, hfun=0x8124000 <cmd_error>) at eval.c:1511
#19 0x08123553 in command_loop_2 () at keyboard.c:1338
#20 0x08181cda in internal_catch (tag=137913897, 
    func=0x8123530 <command_loop_2>, arg=137874713) at eval.c:1247
#21 0x08123e67 in command_loop () at keyboard.c:1317
#22 0x081241db in recursive_edit_1 () at keyboard.c:942
#23 0x08124311 in Frecursive_edit () at keyboard.c:1004
#24 0x08118d8c in main (argc=1, argv=0xbfcfd744) at emacs.c:1728
(gdb) 
The above 
similarly i get many hangs for other key combinations like
C-u C-\ malayalam-inscript RET kra
there k can be many other letter probably those ones which can join with
"a"i.e. a="ോ" for me.
interestingly this crash dont happen with emacs' own keymap. But only
with the xkb's layout.
k is ക but that can also be i: ഗ but not D:അ 
t is ൂ
anyway all time, "a"(ോ) is creating some problems like if ae or qw makes the characters maped
to e and w to display twice. May be a is giving an extra space
considering the letter to be included usually between the two parts of
"ോ"
i am using meera and surprisingly with rachana font, the above specific
bugs are not there.. still it too crashes at some point which i am yet
to track with..
Other than crashes, the display also gets weird at some time for some
random key sequence with malayalam characters
"രോപ്ുോപര്ുേതോപ്തകോര" this is an example of such a thing.
The picture below shows the rendering difference between that in emacs
and in gedit.
   But emacs too makes the rendering perfect at times like when that
   text is selected for yanking. The gedit too is not perfect in rendering the
   above sequence. Avoiding those dotted circles are the thing avoided
   to make perfect rendering for gnome.


Thanks a lott
Shyam K

Re: Rendering Fix progresses, Now Input-Method

[Shyam | ശ്യാം കാരനാട്ട് | Karanattu]

Hi again,
Though the errors reported at the last mail can be adjusted for the
moment, I have a more Serious Problem. I could get the Perfect rendering
in emacs in general But the text area of Po-mode, where it is primarily
needed, is rendering in the old type. I didnt yet get why that buffer alone is
not taking the changes. Tried changing the fontset and
composition-function-table but no way out!:( . The strings as entered to
the po-mode after finalising with C-c C-c is rendered perfectly, but it
fails where it matters i.e at the text edit area, the buffer named
"*po-filename*" where po-filename is the name of the po-file being
edited. The real joy of getting emacs with malayalam rendering comes
when it helps in serious translation works like that of the gnu.org
pages . Sadly the very same got affected :( All other buffers i have
checked so far works perfectly.  

Thanks a lott
Shyam K

 

Re: Rendering Fix progresses, Now Input-Method

[Kenichi Handa]

In article <871vybxja1.fsf@swathanthran.in>, mail@swathanthran.in (Shyam | ശ്യാം കാരനാട്ട്  | Karanattu) writes:

> Though the errors reported at the last mail can be adjusted for the
> moment, I have a more Serious Problem. I could get the Perfect rendering
> in emacs in general But the text area of Po-mode, where it is primarily
> needed, is rendering in the old type.

What do you mean by "old type"?  No composing?  Please show
me the result of C-u C-x = on a Malayalam character that
should be composed with the following characters.  What is
the value of auto-composition-mode in that buffer?

> I didnt yet get why that buffer alone is
> not taking the changes. Tried changing the fontset and
> composition-function-table but no way out!:( . The strings as entered to
> the po-mode after finalising with C-c C-c is rendered perfectly, but it
> fails where it matters i.e at the text edit area, the buffer named
> "*po-filename*" where po-filename is the name of the po-file being
> edited.

As I've never used po-mode, I don't know how to reproduce
what you did.  Please show me the exact recipe to reproduce
that problem.

---
Kenichi Handa
handa@ni.aist.go.jp