Re: [vile] system-mapped-chars

[Thomas Dickey]

On Fri, Dec 26, 2014 at 02:54:35PM -0500, Wayne Cuddy wrote:
> On Tue, Dec 23, 2014 at 08:16:11PM -0500, Thomas Dickey wrote:
> > On Tue, Dec 23, 2014 at 01:02:23PM -0500, Wayne Cuddy wrote:
> > > Can system-defined maps be adjusted after compile time? It doesn't
> > > seem so.
> > 
> > no - it's built-in...
> > 
> > Very briefly, it uses the terminfo/termcap/whatever database for
> > home/end, and adds in - if asked - the xterm function-keys.
> > 
> > The latter isn't much used (but I should rework it, since I've
> > incorporated much of that into ncurses as extended capabilities).  My
> > main use for it is to ensure that special keys come in as a special key
> > rather than individual bytes.
> >  
> > but I digress...
> > 
> > > I've found that HOME/END keypad keys don't always map to #H/#E
> > > respectively.
> > 
> > That really sounds more like a mismatch between terminal emulator
> > and terminal database.  (If it's some program setting TERM to "xterm"
> > but not really xterm, I've been there a lot).
> > 
> > Given some more details, I may be able to offer advice.
> >  
> > > I've been able to get around this with additional maps for
> > > terminal-specific sequences in the vilerc file but it would be nice if
> > > there was an easy way to add to the system maps.
> > > 
> > > Wayne
> > > 
> 
> Sorry, took me a bit of research, this was actually an issue for a
> co-worker. I don't use the HOME/END or really any of the keypad keys, to
> me that's what makes vi-style editors so efficient.
> 
> Aside from Linux we also use HP-UX at the shop. We run HP's OEM xterm as
> well as rxvt-unicode that we built ourselves. We don't really have a
> preference except that rxvt tends to be lighter with respect to network
> I/O between itself and the X-server. This normally isn't noticed but we
> use a VNC-style terminal server environment and high output/scrolling
> can have a noticeable effect on the VNC/X host. But this is maybe an
> xterm discussion for another day.

yes... rxvt doesn't send all of the screen updates.  I added a feature
which can make xterm do part of that, a few years ago, to get discussion.
 
> Vile is linked with HP's OEM curses (xcurses) library. Surprisingly the
> HOME/END keys work as expected in rxvt. In Xterm, either running locally
> on Linux while ssh'd into an HP host or running HP's xterm via X when
> the HOME/END keys are used they produce ^[OH and ^[OE respectively.

Short: the problem can be fixed by correcting the terminal description on
the HP-UX machine.  I still have access to at least one HP-UX server, and
can see this description running "infocmp xterm":

#       Reconstructed via infocmp from file: /usr/share/lib/terminfo/x/xterm
xterm|vs100|xterm terminal emulator,
        am, km, xenl,
        cols#80, lines#65,
        bel=^G, blink=@, bold=\E[1m, civis=\E[?25I,
        clear=\E[H\E[2J, cnorm=\E[?25h, cr=\r,
        csr=\E[%i%p1%d;%p2%dr, cub=\E[%p1%dD, cub1=\b,
        cud=\E[%p1%dB, cud1=\E[B, cuf=\E[%p1%dC, cuf1=\E[C,
        cup=\E[%i%p1%d;%p2%dH, cuu=\E[%p1%dA, cuu1=\E[A,
        cvvis=\E[?25h, dch=\E[%p1%dP, dch1=\E[P, dl=\E[%p1%dM,
        dl1=\E[M, ed=\E[J, el=\E[K, home=\E[1;1H, ht=\t,
        hts=\EH, ich=\E[%p1%d@, ich1=\E[@, il=\E[%p1%dL,
        il1=\E[L, ind=\n, kbs=\177, kcub1=\EOD, kcud1=\EOB,
        kcuf1=\EOC, kcuu1=\EOA, kf1=\EOP, kf10=\E[21~,
        kf11=\E[23~, kf12=\E[24~, kf2=\EOQ, kf3=\EOR,
        kf4=\EOS, kf5=\E[15~, kf6=\E[17~, kf7=\E[18~,
        kf8=\E[19~, kf9=\E[20~, knp=\E[6~, kpp=\E[5~, rc=\E8,
        rev=\E[7m, ri=\EM, rmkx=\E[?1l, rmso=\E[m,
        rs1=\E>\E[1;3;4;5;6l\E[?7h\E[m\E[r\E[2J\E[H, rs2=@,
        sc=\E7,
        sgr=\E[%?%p1%t;7%;%?%p2%t;4%;%?%p3%t;7%;%?%p4%t;5%;%?%p6%t;1%;m,
        sgr0=\E[m, smkx=\E[?1h, smso=\E[7m, tbc=\E[3g,

I'd fix that by 
        infocmp xterm >foo
        --edit foo, adding (a line with a tab at the front):
                khome=\EOH, kend=\EOE,
        sudo tic foo

Long: here's how we got to this point...

The problem is that different terminal emulators use different character
sequences for some of the special keys - Home/End are particular problems.
I've been responsible for some of that, though it was about ten years ago.

hmm - my faq here isn't that clear:
        http://invisible-island.net/ncurses/ncurses.faq.html#home_end_keys
so I'll provide an explanation here, and make a to-do to improve that.
But see these, where I've done a better explanation:
        http://invisible-island.net/xterm/xterm.faq.html#xterm_pageup
        http://invisible-island.net/xterm/xterm.faq.html#xterm_pc_style

There are a couple of issues:

a) If you want to pretend that xterm is a vt220, there is no such thing as
   Home/End.  The corresponding positions on a vt220 keyboard are Select/Find

   If there's no tinkering with the translations resource, the sunKeyboard
   resource is the important one.

b) Some of the terminfo's which I made use the vt220 convention (and is what
   I'm referring to as "ten years ago".

c) If the terminfo doesn't match the terminal behavior, the application won't
   recognize the keys.

d) Here's a script which I ran to show the differences

   Note that xterm-r5 and xterm-r6 assume vt220 keyboard -- using different
   terminfo names -- while the others do not.  Eventually I got it right,
   but was reluctant to redo longstanding descriptions:
--------(script)---------------------------------------------------------------
#!/bin/sh
for term in xterm-r5 xterm-r6 xterm-xfree86 xterm-new xterm-vt220 rxvt
do
echo "** $term"
case $term in
xterm-r6)
        HOME=kfnd
        END=kslt
        ;;
*)
        HOME=khome
        END=kend
        ;;
esac
printf "$HOME:";tput -T$term $HOME |cat -v; echo
printf "$END:";tput -T$term $END |cat -v; echo
done
--------(output)---------------------------------------------------------------
** xterm-r5
khome:^[[1~
kend:^[[4~
** xterm-r6
kfnd:^[[1~
kslt:^[[4~
** xterm-xfree86
khome:^[OH
kend:^[OF
** xterm-new
khome:^[OH
kend:^[OF
** xterm-vt220
khome:^[[1~
kend:^[[4~
** rxvt
khome:^[[7~
kend:^[[8~
-------------------------------------------------------------------------------
rxvt isn't a problem because I wasn't there improving its keyboard
configurabilty (essentially it's hardcoded to one choice - xterm has
several features which multiply to a lot of choices, and the blunder
that I noted in the last link above was due to overlooking a customization
that I was testing...).
 
> On top of that the user wanted the HOME/END keys to operate on the
> current line instead of the entire file. So here are the vilerc
> entries I created.
> 
> ----------------------------------------------------------------------
> map #H 0
> map #E $
> map! #H ^[0i
> map! #E ^[A
> 
> ; Handle HP Xterm

Actually, "recent" HP Xterm is a version from my development (at least
that's what is installed on the machine I can inspect).  You can check
this by typing
        xterm -v
to see if it prints a version message.  (If it simply runs xterm, then
it's _very_ old).

It is actually 11 years old.  But (looking at the terminal description
which I quoted above), the terminfo database does not appear to have
been updated past X11R5, about 20 years ago.

> map ^[OH 0
> map ^[OF $
> map! ^[OF ^[A
> map! ^[OH ^[0i
> ----------------------------------------------------------------------
> 
> The above tends to work quite well so perhaps this is sufficient. I
> was just surprised that the HP Xterm didn't produce #H/#E forcing me
> to map the terminal escapes.
> 
> Thanks again,
> Wayne
> 
> _______________________________________________
> vile mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/vile

-- 
Thomas E. Dickey <address@hidden>
http://invisible-island.net
ftp://invisible-island.net

signature.asc
Description: Digital signature