terminfo(5) manual page describes a set of capabilities wnum (maximum number of definable windows), cwin (define a window), wingo (go to window), wind (resize current window), but only one terminal definition in terminfo master file uses any of them (tvi9065; it sets wnum=0).
Do any hardware terminals, or terminal emulators, exist that support these capabilities?
Short: few terminals have provided these features. Good luck on finding one.
Long: determining if a terminal supports the windowing features is misleading, because the features which are most-used from terminfo are those used in curses. But it's a (weak) clue.
Both ncurses and the AT&T SVr4 terminal descriptions include a few.
Just considering these capabilities listed in terminfo(5):
maximum_windows wnum MW maximum number of
definable windows
create_window cwin CW define a window #1
from #2,#3 to #4,#5
goto_window wingo WG go to window #1
set_window wind wi current window is
lines #1-#2 cols
#3-#4
ncurses lists only a few using wind (none using the others, since stating zero windows is redundant):
Data General d412-unix, etc. (see manual for d413)
Datapoint dp8242
SCO (e.g., OpenServer) sco-ansi
Those particular entries were added a while back (nothing recent):
# 10.1.14 (Sat Nov 22 19:59:03 EST 1997)
# * add vt220-js, pilot, rbcomm, datapoint entries from esr's 27-jun-97
# version.
# * add hds200 description (Walter Skorski)
# * add EMX 0.9b descriptions
# * correct rmso/smso capabilities in wy30-mc and wy50-mc (Daniel Weaver)
# * rename xhpterm back to hpterm.
# 1998/9/26
# * format most %'char' sequences to %{number}
# * adapt IBM AIX 3.2.5 terminfo - T.Dickey
# * merge Data General terminfo from Hasufin <[email protected]> - TD
# 2002-05-25
# * add kf13-kf48 strings to cons25w -TD
# * add pcvt25-color entry -TD
# * changed a few /usr/lib/tabset -> /usr/share/tabset.
# * improve some features of scoansi entry based on SCO's version -TD
# * add scoansi-new entry corresponding to OpenServer 5.0.6
There also is a comment on hds200 which indicates that wind was possible, but conflicted with another use.
The AT&T terminal descriptions likewise had few that used windowing. The SCO terminfo which was the source for much of ncurses in 1995-1996 had a commented-out wind in the description of the Concept AVT:
# Info:
# Concept AVT with status line. We get the status line using the
# "Background status line" feature of the terminal. We swipe the
# first line of memory in window 2 for the status line, keeping
# 191 lines of memory and 24 screen lines for regular use.
# The first line is used instead of the last so that this works
# on both 4 and 8 page AVT's. (Note the lm#191 or 192 - this
# assumes an 8 page AVT but lm isn't currently used anywhere.)
#
avt+s|concept avt status line changes,
is3=\E[2w\E[2!w\E[1;1;1;80w\E[H\E[2*w\E[1!w\E2\r\n,
tsl=\E[2;1!w\E[;%p1%dH\E[2K, fsl=\E[1;1!w, eslok, hs,
dsl=\E[0*w, lm#191, smcup=\E[2;25w\E2\r, rmcup=\E[2w\E2\r\n,
.wind=\E[%i%p1%{1}%+%d;%p2%d;%p3%{01}%+%d;%p4%{01}%+%dw
A comment on one which I had from an OSF/1 machine says this:
# EXECUTION ENVIRONMENT:
#
# This entry does not use any of the fancy windowing stuff of the
# 2626. Indeed, terminfo does not yet handle such stuff. Since
# changing any window clears memory, it is probably not possible to
# use this for screen opt. ed is incredibly slow most of the time.
# It may due to the exact padding.
#
# Since the terminal uses xoff/xon this is intended only for cost
# computation, so that the terminal will prefer el or even dl1 which
# is probably faster! \ED\EJ\EC is also being used -
# apparently ed is only extra slow on the last line of the window.
# The padding probably should be changed.
hp2626|hp2626a|hp2626p|2626|2626a|2626p|2626A|2626P|hp 2626,
but (see manual) on closer inspection you probably would notice that each of these terminals has a different feature which wind does not entirely address (it's too broad a range of possibilities, and too few terminals that provided window-like features to justify their use in screen-optimization).
