mlterm-dev-en — mlterm mailing list(English)

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Mike F. <mf...@su...>]

MINAMI Hirokazu <mi...@mi...> さんは書きました:

> On Sun, 2005-01-09 at 16:38 +0100, Mike FABIAN wrote:
>> I think yes because I am using the configure option --with-libtdl.
>> I enable this configure option only for SuSE Linux > 9.0 though:
>> 
>>     %if %suse_version > 900	    
>>                 --with-libltdl \
>>     %endif
>> 
>> for SuSE Linux <= 9.0 it doesn't work with this option, for
>> SuSE Linux > 9.0 it doesn't work without this option.  I have not yet
>> investigated why, I don't clearly understand this stuff yet.
>
> I'm not yet sure why you have to use libtool on SuSE >9.0.
> What was not working, compiling or dynamic loading?

compiling worked but dynamic loading failed on SuSE 9.1 and SuSE 9.2.
For example the scrollbar plugins didn't work anymore without that
configure option. But I really don't know why.

-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[MINAMI H. <mi...@mi...>]

On Mon, 2005-01-10 at 20:31 +0900, Seiichi SATO wrote:

> TD's xterm(197):
> - saved/restored .... no
> - shared between normal/alternate screen .... yes
> 
> rxvt 2.6.4:
> - saved/restored .... no
> - shared between normal/alternate screen .... no
> 
> mlterm (2.9.1 pre/cvs-1.806)
> - saved/restored .... no
> - shared between normal/alternate screen .... yes
> 
> In the implementation of cursor state, it seems to be compatible
> with TD's xterm.

Thanks for checking.

The rxvt's behavior may be desireble for some purposes
since tere is no way to know current cursor visibility from 
an application running on a terminal.

However, it should be almost whenever enough to
simply revert cursor to the normal state after drawing finished.
So I'd like to leep current implementation for a while.

--
minami

 

 

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[MINAMI H. <mi...@mi...>]

On Sun, 2005-01-09 at 16:38 +0100, Mike FABIAN wrote:
> I think yes because I am using the configure option --with-libtdl.
> I enable this configure option only for SuSE Linux > 9.0 though:
> 
>     %if %suse_version > 900	    
>                 --with-libltdl \
>     %endif
> 
> for SuSE Linux <= 9.0 it doesn't work with this option, for
> SuSE Linux > 9.0 it doesn't work without this option.  I have not yet
> investigated why, I don't clearly understand this stuff yet.

I'm not yet sure why you have to use libtool on SuSE >9.0.
What was not working, compiling or dynamic loading?

--
minami

Re: [Mlterm-dev-en] Shift+space=IM_HOTKEY and --im=m17nlib:ja:anthy

[Mike F. <mf...@su...>]

Seiichi SATO <ss...@sh...> さんは書きました:

> On Fri, Jan 14, 2005 at 12:33:32PM +0100,
>  Mike FABIAN wrote:
>
>> Why is there a difference, i.e. why is it necessary to configure
>> this key binding for --im=m17nlib:ja:anthy but not for --im=uim?
>
> Yes, IM_HOTKEY is required only from --im=m17nlib and --im=kbd.
> When using uim it has no effect, because the hotkey is defined in
> libuim. (mlterm cannot define it.)

Ah, I see. Thank you!

-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Seiichi S. <ss...@sh...>]

Mike,

I fixed kik_is_dl_module() in kik_dlfcn_ltdl.c. (rev 1.4)
If plugin directory contains .a and .la files, IM plugin now
works fine with libltdl.

Thanks for your report.

-- 
Seiichi

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Seiichi S. <ss...@sh...>]

On Fri, Jan 14, 2005 at 12:20:21PM +0900,
 MINAMI Hirokazu wrote:

> The optimization, unify all space's fg color to be default one,
> seems to be no longer useful for me.

I think so too.

> I'm going to drop it and introduce another optimization,
> assume we can draw blank characters in whatever color we want,
> into text rendering subsystem.

That sounds good.

Some people said the scrolling speed of mlterm is slow. To solve
this proble, we need to replace --enable-optimize-redrawing with
another optimization.

Designing new rendering model is not easy. But I hope it will be
placed top of TODO for mlterm-3.x. Of course, I will cooperate
with you.

-- 
Seiichi

Re: [Mlterm-dev-en] Shift+space=IM_HOTKEY and --im=m17nlib:ja:anthy

[Seiichi S. <ss...@sh...>]

On Fri, Jan 14, 2005 at 12:33:32PM +0100,
 Mike FABIAN wrote:

> Why is there a difference, i.e. why is it necessary to configure
> this key binding for --im=m17nlib:ja:anthy but not for --im=uim?

Yes, IM_HOTKEY is required only from --im=m17nlib and --im=kbd.
When using uim it has no effect, because the hotkey is defined in
libuim. (mlterm cannot define it.)

-- 
Seiichi

[Mlterm-dev-en] Shift+space=IM_HOTKEY and --im=m17nlib:ja:anthy

[Mike F. <mf...@su...>]

When starting

    mlterm --im=uim:anthy

I can activate the input via uim by typing Shift+space although
I have *not* configured a key binding for IM_HOTKEY in ~/.mlterm/key.

But when starting 

    mlterm --im=m17nlib:ja:anthy

this doesn't work without configuring

    Shift+space=IM_HOTKEY

in !~/.mlterm/key .

Why is there a difference, i.e. why is it necessary to configure
this key binding for --im=m17nlib:ja:anthy but not for --im=uim?


-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

[Mlterm-dev-en] Invisible cursor with python curses module and start_color

[MINAMI H. <mi...@mi...>]

On Wed, Jan 12, 2005 at 08:58:19PM -0500, Ian Ward wrote:
> On Thu, 2005-13-01 at 05:02 +0900, Seiichi SATO wrote:
> > I made a little change to the patch.
> > Please try it again.
> 
> Perfect.  Removing that optimization fixed it.  Perhaps the problem
> could also be solved by just temporarily disabling that optimization
> when clearing the screen.

The optimization, unify all space's fg color to be default one,
seems to be no longer useful for me.

I'm going to drop it and introduce another optimization,
assume we can draw blank characters in whatever color we want, 
into text rendering subsystem.

Comments?


P.S. in unicode, 0x20 is not the only blank character.
It may better to apply same optimization anyway
for other characters like 0x3000<IDEOGRAPHIC SPACE>, 0x2002<EN SPACE> etc.
Does anyone have a list of them?

--
minami

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Ian W. <ia...@ex...>]

On Thu, 2005-13-01 at 05:02 +0900, Seiichi SATO wrote:
> I made a little change to the patch.
> Please try it again.

Perfect.  Removing that optimization fixed it.  Perhaps the problem
could also be solved by just temporarily disabling that optimization
when clearing the screen.

Thanks!

Ian

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Seiichi S. <ss...@sh...>]

Hi Ian,

On Tue, Jan 11, 2005 at 10:12:26PM -0500,
 Ian Ward wrote:

> > > I tried the patch on 2.9.1 and the cvs version, but I still can't see a
> > > cursor in my small test case.  I couldn't find a config file setting for
> > > the cursor colour.  Any Ideas?

I made a little change to the patch.
Please try it again.

Regards,

-- 
Seiichi

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Ian W. <ia...@ex...>]

On Tue, 2005-11-01 at 22:42 +0900, MINAMI Hirokazu wrote:
> > I tried the patch on 2.9.1 and the cvs version, but I still can't see a
> > cursor in my small test case.  I couldn't find a config file setting for
> > the cursor colour.  Any Ideas?
> 
> Hmm, I could see a cursor in "white" after the message "No cursor".
> Could you try to display some colored chars
> ex)
>  tput cup 2 0;echo -e "\033[31mcolored_cursor";tput cup 0 0
> and see what happened if you move a cursor into them?

Yes, the cursor turns red when I move it over the red text.  Also, if I
move the cursor over the "No cursor" text in the curses program it
appears as a white cursor.  It seems that if I put it anywhere else on
the screen it is invisible (black on black?), including when I put the
cursor on the space in between "No" and "cursor".

xterm and rxvt seem to default to white on black for the entire screen
in my test case. gnome-terminal uses light gray on black. 

Ian

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[MINAMI H. <mi...@mi...>]

On Mon, 2005-01-10 at 09:14 -0500, Ian Ward wrote:
> On Tue, 2005-04-01 at 11:49 +0900, MINAMI Hirokazu wrote:
> > I've prepared a patch for the xterm-like scheme (attached).
> > i.e. make cursor's fg/bg color be taken from the character under cursor.
> 
> I tried the patch on 2.9.1 and the cvs version, but I still can't see a
> cursor in my small test case.  I couldn't find a config file setting for
> the cursor colour.  Any Ideas?

Hmm, I could see a cursor in "white" after the message "No cursor".
Could you try to display some colored chars
ex)
 tput cup 2 0;echo -e "\033[31mcolored_cursor";tput cup 0 0
and see what happened if you move a cursor into them?


Though You can specify fg/bg color for cursor by adding 
corresponding entries (cursor_fg_color/cursor_bg_color)
to your ~/.mlterm/main, my patch would do nothing in that case
since the color you set, not the color of a character under cursor,
would be always used. 


--
minami

[Mlterm-dev-en] Re: mlterm will freeze if I switch scim-m17n engines

[Mike F. <mf...@su...>]

Zhe Su <jam...@gm...> さんは書きました:

>   Yes, it works, at least for me :-) Thank you very much.
>   Anyone else can help test this patch?

After updating to scim 1.1.1 (CVS snapshot from today which includes
SATO-san's patch), I can switch input methods in mlterm without
problems.

Thank you very much!

> On Sun, 9 Jan 2005 19:31:50 +0900, Seiichi SATO <ss...@sh...> wrote:
>> On Wed, Jan 05, 2005 at 12:59:47AM +0900,
>>  UTUMI Hirosi wrote:
>> 
>> > SATO-san, Minami-san, what do you think about it?
>> 
>> I think that the problem happens in the event dispatcher
>> in scim_x11_frontend.cpp.
>> Could anybody try the attached patch?
>> 
>> --
>> Seiichi
>> 
>> 
>> _______________________________________________
>> scim mailing list
>> sc...@li...
>> http://lists.freedesktop.org/mailman/listinfo/scim
>> 
>> 
>> 
>>
> _______________________________________________
> scim mailing list
> sc...@li...
> http://lists.freedesktop.org/mailman/listinfo/scim
>

-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Ian W. <ia...@ex...>]

On Tue, 2005-04-01 at 11:49 +0900, MINAMI Hirokazu wrote:
> On Sun, 2005-01-02 at 01:16 -0500, Ian Ward wrote:
> 
> > xterm always flips the foreground and background colors of the character
> > the cursor is on (white on blue becomes blue on white etc.)  rxvt uses
> > black on white if the background of the character the cursor is on is
> > black, otherwise it uses white on black.
> 
> I've prepared a patch for the xterm-like scheme (attached).
> i.e. make cursor's fg/bg color be taken from the character under cursor.

I tried the patch on 2.9.1 and the cvs version, but I still can't see a
cursor in my small test case.  I couldn't find a config file setting for
the cursor colour.  Any Ideas?

-------------------------
#!/usr/bin/python

import curses
import sys
s = curses.initscr()
curses.start_color()
s.addstr("No cursor")
s.refresh()
sys.stdin.read(1)
curses.endwin()

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Seiichi S. <ss...@sh...>]

On Sun, Jan 09, 2005 at 08:00:02PM +0900,
 Seiichi SATO wrote:

> On Sun, Jan 02, 2005 at 09:58:35AM +0900,
>  MINAMI Hirokazu wrote:
> 
> > BTW, cursor state(invis/norm) currently doesn't save/restored
> >  and are sheared between normal/alternate screen.
> > Are they intentional ?> Sato san
> 
> Those are not intentional.
> I submitted:
> 
> <https://sourceforge.net/tracker/index.php?func=detail&aid=1098851&group_id=40891&atid=429201>

TD's xterm(197):
- saved/restored .... no
- shared between normal/alternate screen .... yes

rxvt 2.6.4:
- saved/restored .... no
- shared between normal/alternate screen .... no

mlterm (2.9.1 pre/cvs-1.806)
- saved/restored .... no
- shared between normal/alternate screen .... yes

In the implementation of cursor state, it seems to be compatible
with TD's xterm.

-- 
Seiichi

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Mike F. <mf...@su...>]

MINAMI Hirokazu <mi...@mi...> さんは書きました:

> On Sun, 2005-01-09 at 14:09 +0100, Mike FABIAN wrote:
>> Seiichi SATO <ss...@sh...> さんは書きました:
>> > When loading plugins, mlconfig checks plugin extensions such as
>> > so, sl and dll using kik_dl_is_module() in kik_dlfcn_*.c.
>> > Therefore, mlconfig skips loading .a and/or .la files.
>> >
>> > Is kik_dl_is_module() correctly working on your environment?
>> 
>> Loading of the plugins worked fine and the plugins also worked.
>> There was only a problem in the GUI of mlconfig. But the patch
>> by MINAMI Hirokazu seems to fix it.
>
> I've assumed that kik_dlfcn_ltdl.c is selected on your environment
> and kik_dl_is_module() is checking nothing.
> Is this correct? > Mike

I think yes because I am using the configure option --with-libtdl.
I enable this configure option only for SuSE Linux > 9.0 though:

    %if %suse_version > 900	    
                --with-libltdl \
    %endif

for SuSE Linux <= 9.0 it doesn't work with this option, for
SuSE Linux > 9.0 it doesn't work without this option.  I have not yet
investigated why, I don't clearly understand this stuff yet.

> In my understanding, "ltdl" is to use libtool for emulating
> dynamic loading on a system that do not support such feature.
>
> Do we really have to install *.a on such systems?
> Shouldn't they be statically linked and not required at run-time?
> #I'm not sure how we can enumerate plug-ins in that case, though.

I tried deleting the *.a and it still works (on SuSE 9.2 where
I currently use --with-libltdl).

-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Seiichi S. <ss...@sh...>]

On Sun, Jan 09, 2005 at 10:50:55PM +0900,
 MINAMI Hirokazu wrote:

> On Sun, 2005-01-09 at 14:09 +0100, Mike FABIAN wrote:
> > Seiichi SATO <ss...@sh...> さんは書きました:
> > > When loading plugins, mlconfig checks plugin extensions such as
> > > so, sl and dll using kik_dl_is_module() in kik_dlfcn_*.c.
> > > Therefore, mlconfig skips loading .a and/or .la files.
> > >
> > > Is kik_dl_is_module() correctly working on your environment?
> > 
> > Loading of the plugins worked fine and the plugins also worked.
> > There was only a problem in the GUI of mlconfig. But the patch
> > by MINAMI Hirokazu seems to fix it.
> 
> I've assumed that kik_dlfcn_ltdl.c is selected on your environment
> and kik_dl_is_module() is checking nothing.

Oops. You are right. kik_dl_is_module() in kik_dlfcn_ltdl.c
do nothing. It is a bug.

> Do we really have to install *.a on such systems?
> Shouldn't they be statically linked and not required at run-time?
> #I'm not sure how we can enumerate plug-ins in that case, though.

If ltdl does not required .a, it should not installed.
But I don't know that whether it is required or not.

Anyway, I think Minami-san's patch (is_registerd()) is better than
checking pluggable module by kik_dl_is_module().

--
Seiichi

[Mlterm-dev-en] Re: [SCIM] Re: mlterm will freeze if I switch scim-m17n engines

[Zhe Su <jam...@gm...>]

Hi,
  Yes, it works, at least for me :-) Thank you very much.
  Anyone else can help test this patch?

Regards
James Su


On Sun, 9 Jan 2005 19:31:50 +0900, Seiichi SATO <ss...@sh...> wrote:
> On Wed, Jan 05, 2005 at 12:59:47AM +0900,
>  UTUMI Hirosi wrote:
> 
> > SATO-san, Minami-san, what do you think about it?
> 
> I think that the problem happens in the event dispatcher
> in scim_x11_frontend.cpp.
> Could anybody try the attached patch?
> 
> --
> Seiichi
> 
> 
> _______________________________________________
> scim mailing list
> sc...@li...
> http://lists.freedesktop.org/mailman/listinfo/scim
> 
> 
> 
>

[Mlterm-dev-en] Re: [SCIM] Re: mlterm will freeze if I switch scim-m17n engines

[UTUMI H. <utu...@ya...>]

Hi,

--- Seiichi SATO <ss...@sh...> wrote
> I think that the problem happens in the event dispatcher
> in scim_x11_frontend.cpp.
> Could anybody try the attached patch?

Great! It's fixed now.
I tested it on scim-1.1.0.

Thank you for the patch and your great mlterm!

Hirosi


__________________________________
Do You Yahoo!?
Upgrade Your Life
http://bb.yahoo.co.jp/

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[MINAMI H. <mi...@mi...>]

On Tue, 2005-01-04 at 01:21 -0500, Ian Ward wrote:
> > Just curious: what function are you going to add? 
> > I'm assuming that the knowledge of encoding should not
> > be neccesary to simply display some strings.
> 
> I need to know the encoding for things like word wrapping and cursor
> movement in edit boxes (sometimes the cursor should move 2 spaces).  I
> will also have to know how the terminal will behave when I send it
> combining characters and BiDi text.  I don't support BiDi yet -- That
> will be an interesting challenge.

As far as, I know there are no standard which defines
"proper" behavior of a terminal for BiDi/full-width/combining etc.

For mlterm, there are many options to do what is expected: 
--comb
--dyncomb
--multicol
--ac
--bi

Since most of these options are dynamically configurable,
it's very difficult to know what mlterm will do 
for a application running on it.

Maybe you can refer the way of mined (http://towo.net/mined/,
an unicode capable text editor runs on a terminal).

If mlterm was the only terminal in the world, we could take another 
approach. i.e. change the mlterm's configuration as your library want.
But that won't work other terminal emulators...

--
minami

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[MINAMI H. <mi...@mi...>]

On Sun, 2005-01-09 at 14:09 +0100, Mike FABIAN wrote:
> Seiichi SATO <ss...@sh...> さんは書きました:
> > When loading plugins, mlconfig checks plugin extensions such as
> > so, sl and dll using kik_dl_is_module() in kik_dlfcn_*.c.
> > Therefore, mlconfig skips loading .a and/or .la files.
> >
> > Is kik_dl_is_module() correctly working on your environment?
> 
> Loading of the plugins worked fine and the plugins also worked.
> There was only a problem in the GUI of mlconfig. But the patch
> by MINAMI Hirokazu seems to fix it.

I've assumed that kik_dlfcn_ltdl.c is selected on your environment
and kik_dl_is_module() is checking nothing.
Is this correct? > Mike


In my understanding, "ltdl" is to use libtool for emulating
dynamic loading on a system that do not support such feature.

Do we really have to install *.a on such systems?
Shouldn't they be statically linked and not required at run-time?
#I'm not sure how we can enumerate plug-ins in that case, though.

> Sato-san

--
minami

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Mike F. <mf...@su...>]

Seiichi SATO <ss...@sh...> さんは書きました:

> On Wed, Jan 05, 2005 at 07:28:59PM +0100,
>  Mike FABIAN wrote:
>
>> 1st crash:
>> ==========
>> 
>> If the plugin directory contains
>> 
>>     libim-kbd.so
>>     libim-kbd.a
>>     libim-kbd.la
>>     libim-uim.so
>>     libim-uim.a
>>     libim-uim.la
>> 
>> and not only the .so files, mlconfig will think the keyboard and uim
>> plugins exist 3 times and will paint 3 buttons for each. See this
>
> When loading plugins, mlconfig checks plugin extensions such as
> so, sl and dll using kik_dl_is_module() in kik_dlfcn_*.c.
> Therefore, mlconfig skips loading .a and/or .la files.
>
> Is kik_dl_is_module() correctly working on your environment?

Loading of the plugins worked fine and the plugins also worked.
There was only a problem in the GUI of mlconfig. But the patch
by MINAMI Hirokazu seems to fix it.

-- 
Mike FABIAN   <mf...@su...>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。

Re: [Mlterm-dev-en] mlconfig crashes with current CVS head

[Seiichi S. <ss...@sh...>]

On Wed, Jan 05, 2005 at 07:28:59PM +0100,
 Mike FABIAN wrote:

> 1st crash:
> ==========
> 
> If the plugin directory contains
> 
>     libim-kbd.so
>     libim-kbd.a
>     libim-kbd.la
>     libim-uim.so
>     libim-uim.a
>     libim-uim.la
> 
> and not only the .so files, mlconfig will think the keyboard and uim
> plugins exist 3 times and will paint 3 buttons for each. See this

When loading plugins, mlconfig checks plugin extensions such as
so, sl and dll using kik_dl_is_module() in kik_dlfcn_*.c.
Therefore, mlconfig skips loading .a and/or .la files.

Is kik_dl_is_module() correctly working on your environment?

-- 
Seiichi

Re: [Mlterm-dev-en] Invisible cursor with python curses module and start_color

[Seiichi S. <ss...@sh...>]

On Sun, Jan 02, 2005 at 09:58:35AM +0900,
 MINAMI Hirokazu wrote:

> BTW, cursor state(invis/norm) currently doesn't save/restored
>  and are sheared between normal/alternate screen.
> Are they intentional ?> Sato san

Those are not intentional.
I submitted:

<https://sourceforge.net/tracker/index.php?func=detail&aid=1098851&group_id=40891&atid=429201>

-- 
Seiichi