Re: terminfo [Re: console enhancements: mouse events etc]

[Thomas Wolff <towo at towo dot net>]

On 24.02.2010 01:56, Charles Wilson wrote:
> Thomas Wolff wrote:
>    
> > Charles Wilson wrote:
> >      
> > > Thomas Wolff wrote:
> > >
> > >        
> > > > Actually, I just remember again that I though I should change the
> > > > terminfo entry too. Just - where's the source to patch?
> > > >          
> > > So, send me patches against terminfo.src from that -src tarball, and
> > > once we've got it figured out, I'll push it upstream to the ncurses
> > > maintainer.
> > >
> > >
> > >        
> > Sorry for the slightly late response. Attached is a small patch.
> > Two notes:
> > * I used the occasion to add PC graphics mode to the linux console, too,
> > which have always been missing there.
> >      
> I'm not sure about this.  The only situation where you'd be using the
> cygwin terminfo database while interacting with a linux console is if
> you were running a cygwin shell, under WINE, from a console mode linux
> login.  Or maybe if you were remotely accessing a cygwin box from a
> console mode linux session.  However, in both cases, the terminal
> description would DIFFER from the one the actual linux box, the one
> you'd be sitting in front of, uses.  This sounds like trouble to me.
>
> I think the better approach -- doomed to rejection, unfortunately (*) --
> is to submit this change directly upstream to ncurses (Thomas Dickey).
> If it's picked up there, then it will filter down thru the linux
> distributions, AND we'll get it.
>
> (*) Thomas (Dickey) has a policy of accepting terminal description
> changes ONLY from the maintainers of the emulator/console code: e.g.
> konsole from the KDE devs, linux from lkml only, etc.
>    
Quite understandable. However, I already sent a bug report to 
linux-kernel@vger.kernel.org some months ago and there doesn't seem to 
be anyone behind that mail address who would even care to respond..., so
> He's unlikely to accept a patch to the linux terminfo description from
> us. But, give it a try anyway...
>    
so I had just hoped to get this upstream this way. Anyway, my updated 
patch (attached) is now without the linux detail.

> > * I patched "cygwinDBG" only for now, because if the patch goes
> > upstream, the new VT100 graphics mode will not be available for remote
> > login from an older cygwin console for a while. Feel free to modify the
> > entry "cygwin" accordingly if you feel confident with it.
> >      
> No, I think it would be best to update cygwin itself (and maybe
> cygwinDBG. Anybody use that?)  If we want to add cygwin-old (cygwin-1p5?
> suggestion for a better name?) for backwards compatibility, we can.  But
> the current cygwin terminal description should describe the current
> cygwin terminal capabilities.
>    
OK, updated.

> BTW, does this "fix" 'pstree -G'?
>    
Unfortunately not. I checked pstree and it is obviously quite buggy 
about terminal handling. I also checked the source and it does the wrong 
thing.
There are different methods to embed VT100 graphics codes, for short let 
me call them G1 and G0. What pstree sends is:
G0-on G1-off code G0-off
Obviously terminals handle this differently. xterm and mintty seem to 
still apply VT100 graphics within G0-on...G0-off despite the stray G1-off,
while my cygwin console patch doesn't, it just switches it off when it's 
told "off" by one method and I thought that would be OK.
So there is two options:
- I could amend my patch with a workaround
- I/We could get pstree fixed, probably upstream?


> I'll adapt and release an update relatively soon. ...
> I'm guessing that these terminfo changes need to wait until 1.7.2, right?
>    
Yes, please.

Thomas

Attachment: terminfo-diff.1.7.2-a
Description: Text document

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple