Re: ANN: terminfo-20160924.src.gz

[Mikulas Patocka]

On Sat, 29 Oct 2016, Thomas Dickey wrote:

> On Tue, Oct 25, 2016 at 04:14:26PM +0200, Mikulas Patocka wrote:
> > 
> > 
> > On Mon, 17 Oct 2016, Thomas Dickey wrote:
> > 
> > > On Mon, Oct 10, 2016 at 04:42:11PM +0200, Mikulas Patocka wrote:
> > > > 
> > > > 
> > > > On Sun, 9 Oct 2016, Thomas Dickey wrote:
> > > > 
> > > > > On Wed, Sep 28, 2016 at 02:19:27PM +0200, Mikulas Patocka wrote:
> > > > > > 
> > > > > > 
> > > > > > Hi
> > > > > > 
> > > > > > The linux2.6 terminfo entry still contains the string 
> > > > > > "enacs=\E(B\E)0", 
> > > > > > which makes it unusable with iso-8859-2 locale and most other 
> > > > > > one-byte 
> > > > > > character sets (except iso-8859-1 and utf-8).
> > > > > 
> > > > > From our previous discussion, it wasn't clear to me why you cannot use
> > > > > 
> > > > >       # CONFIGURATION FILE FOR SETUPCON
> > > > > 
> > > > >       # Consult the console-setup(5) manual page.
> > > > > 
> > > > >       ACTIVE_CONSOLES="/dev/tty[1-6]"
> > > > > 
> > > > >       #CHARMAP="UTF-8"
> > > > >       CHARMAP="ISO-8859-2"
> > > > > 
> > > > >       CODESET="Lat2"
> > > > >       FONTFACE="Fixed"
> > > > >       FONTSIZE="8x16"
> > > > > 
> > > > >       VIDEOMODE=
> > > > > 
> > > > >       # The following is an example how to use a braille font
> > > > >       # FONT='lat9w-08.psf.gz brl-8x8.psf'
> > > > 
> > > > I'm using this configuration and this configuration doesn't work with 
> > > > current terminfo.
> > > > 
> > > > You can try it - use the above configuration, run setupcon and then run
> > > > LANG=cs_CZ.iso-8859-2 mc
> > > > - at the bottom line, the label for key 6 should be "Presun" (with 
> > > > caron 
> > > > over "r"), but with current terminfo, the letter r-with-caron is 
> > > > corrupted 
> > > > and a different character is printed.
> > > 
> > > not to argue (as I said, I am investigating), I don't see the problem
> > > when I run this script (see attached screenshot):
> > > 
> > > #!/bin/sh
> > > reset
> > > utf8 on
> > > TERM=linux3.0 \
> > > LANGUAGE=cs_CZ \
> > > LANG=cs_CZ.iso88592 mc
> > > 
> > > where "utf8" is this script
> > > 
> > > #!/bin/sh
> > > # send character-string to enable UTF-8 mode
> > > if test ".$1" = ".off" ; then
> > >   printf '\033%%@'
> > > else
> > >   printf '\033%%G'
> > > fi
> > > 
> > > Having seen disagreements over this (both ways), I'll continue 
> > > investigating
> > > until I can make a table illustrating the tradeoffs.
> > 
> > You took the screenshot in X-windows - but this problem manifests on the 
> > console.
> 
> The screenshot was done using VMware Fusion, not an X-windows application.
> If you see a difference, it's due to the way we executed the test.
> I sent a script to show what I did.
>  
> > You must switch away from the Xserver to Linux console to see the problem. 
> > If I run your script on the console, it obviously doesn't work. See this: 
> > http://artax.karlin.mff.cuni.cz/~mikulas/screenshots/screenshot1.jpg
> > 
> > Your script has these problems:
> > * the "reset" command sends "ESC c", that resets the console mapping to 
> >   iso-8859-1 mode, so it breaks the iso-8859-2 character set
> 
> That's too bad.  There is no possibility that I will accept any patch that
> prevents resetting the terminal (the font is one of 38 items modified, not all
> of those can be reset by other means and replacing that escape with an
> incomplete workaround is something that not many people would want).  If you
> want to make an exception, get a feature added to the Linux console which will
> reset it as you want.

I agree - we don't need to change reset behavior. reset is normally not 
used - so the fact that it resets the charset mapping is not a big deal.

> The suggestion to omit the SCS B from enacs is helpful, but changing the
> behavior of reset is not negotiable.

Yes dropping the SCS B would be sufficient to make iso-8859-2 work.

> > * the "utf8 on" command switches the console to utf-8 mode, but the LANG 
> >   variable is set to iso8859-2 mode, so the national characters are not 
> >   displayed at all
> > 
> > If I remove "reset" and "utf8 on" from your script, on the Linux console, 
> > I see this:
> > http://artax.karlin.mff.cuni.cz/~mikulas/screenshots/screenshot2.jpg
> > - you can see that the iso-8859-2 characters are broken.
> 
> thanks (more data :-)
>  
> > When I downgrade the ncurses-base and ncurses-term packages to version 
> > "6.0+20160625-1", the iso-8859-2 characters are displayed correctly (but 
> > that version has broken "linux3.0", so we should use "linux" instead):
> > http://artax.karlin.mff.cuni.cz/~mikulas/screenshots/screenshot3.jpg
> > 
> > Mikulas
> 
> -- 
> Thomas E. Dickey <address@hidden>
> http://invisible-island.net
> ftp://invisible-island.net

Mikulas