[Gterm-discuss] Re: Gterm VT 100

[Jörg S. <joe...@we...>]

On Fri, 6 Dec 2002 20:39:20 +0800
Zhang Le <ej...@pe...> wrote:

> On Thursday 05 December 2002 22:01, Jörg Sonnenberger wrote:
> > Hi Zhang (Le?),
> Nice to see you here.
> Zhang is my surname and Le is my first name which means "joy" in Chinese.

Right, I wasn't sure. Le = joy was new for me. I am a bit wiser now ;-)

> > nice patch. I tried to merge it with my own changes for clean compilation,
> > the resulting diff against CVS is attached. Hope I haven't forgotten
> > anything. If not, could you attach to sourceforge?
> I carefully modified the original gterm VT parser according to the programming 
> section of VT100 user guide(got from vt100.net).
> My changes include:
> 1.fix inversed BW mode problem
> 2.add charset support
> 3.modify some cursor/scroll function so their  behaver like what the VT100 UG 
> said
> 4.use dynamic buffer instead a fixed one
> 5.other enhancements and bug fixes.
> I use vttest (found on freshmeat.net) to test my modified version so I believe 
> there should be no serious bug in VT100 parser(I hope so;-)). The linux kernel 
> itself does not pass the vttest problem.
> Still, the double width char function(char width switch) , smooth scroll 
> function and other VT100 specific commands are not implemented since I 
> thought they are useless in a soft VT100 enumerator.

I agree.

> >
> >
> > Other changes are moving ASSERT_X and ASSERT_Y from gterm.hpp
> > to use .cpp files to keep the namespace clean and adding #ifndef NDEBUG
> > around the no_operator? defines and calls. Hope this is right. Are they
> > used any more?
> no_operator should be deleted, they are non-standard ESC sequence I found in 
> linux kernel's console.c which implements some vga setting function.
> Since gterm is a vt100 colone, they can be removed.

Done.

> >
> >
> > Also can you reproduce one bug in your code? If I am using plain sh and I
> > start ls or ls -w 80, it displays a really weired screen. After the first
> > file is a line break, after the second there is a lot of whitespace and the
> > first char of the third file is on the same line. This doesn't happen under
> > bash or the original gterm. This isn't introduced by the merge, it works
> > with your version too. If you can't reproduce it, I will try to debug it
> > myself.
> I've tracked down the bug. It seems ash(plain sh) does not initiate tab stops 
> on start so we should reset tab stops in GTerm::reset()

I guessed it's that kind of problem.

> here's my patch against CVS code:
> --- actions.cpp.old     Fri Dec  6 20:19:14 2002
> +++ actions.cpp Fri Dec  6 20:19:17 2002
> @@ -222,6 +222,9 @@
>         }
>         for (i=0; i<height; i++) linenumbers[i] = i;
>         memset(tab_stops, 0, width);
> +    for (i=0; i<width;++i)
> +        if ((i + 1) % 8 ==0)
> +            tab_stops[i] = 1;
>         current_state = GTerm::normal_state;
>  
>         clear_mode_flag(NOEOLWRAP | CURSORAPPMODE | CURSORRELATIVE |
> 

Applied. I changed to for-loop and dropped the if. Thanks, now it works.

> 
> -- 
>                                        Sincerely yours,
>                                                   Zhang Le

Sincerely yours,
Joerg Sonnenberger