keep output visible after exiting with ^c or there is no more data

[justin2004]

resolves #41

[justin2004]

works on my linux machines but i tested this on my macbook and it did not work. so i will need to update the term type.

[justin2004]

just to gather some information i wrote a script to see which term types work (leave the output visible).

maybe we include a list of working term types and only call newterm() when an environment already has TERM set to one of the working term types? otherwise we just use initscr().

n cygwin
n cons25
n cons25-debian
n screen-256color
n screen-256color-bce
n screen-w
n screen-bce
n screen
n screen-s
y sun
y linux
y Eterm
y xterm-vt220
y xterm-r5
y xterm-r6
y xterm-color
y xterm-256color
y xterm-mono
y xterm
y xterm-xfree86
n mach-gnu
n mach-gnu-color
y mach-bold
y mach-color
y mach
y vt100
y vt102
y vt220
n vt52
n dumb
n ansi
y wsvt25
y wsvt25m
y rxvt
y rxvt-basic
y rxvt-unicode
n pcansi
n hurd
n cons30-m
n cons30
y cs10
n citoh-6lpi
n cdc456
n cons43
n cit101e-rv
n c108-w
n cons60l1-m
n cdc752
n cons50-m
n cons50r-m
y cit500
n contel300
n cad68-3
y color_xterm
n cygwinDBG
n cons50l1
n cit101e-n132
y cs10-w
n c108-rv-4p
n citoh-prop
n cit101
n cons60r-m
n cit101e-n
n cci
n c108-rv
n cons60l1
broke my script citoh-pica

[tenox7]

Thank you. I would definitely want to stay away from coding for different terminal types in the app. This is the role of ncurses do it for you. I think perhaps some specific capabilities need to be tested or modes set instead?

[tenox7]

I may be completely mistaken but upon further investigation I have determined that this problem has absolutely nothing to do with initscr() vs newterm().

Instead the problem is with xterm (and other terminal) "alternate screen" switching.

This can be worked around in code by implementing an ugly hack from dialog(1) that disables processing enter/exit alt screen. I actually have working solution but currently cannot publish due to incompatible licensing.

Another method is simply to use a different, but still compatible terminal definition, eg

seq 10 | TERM=vt100 ttyplot

seem to work fine on xterm-ish terminals.

[justin2004]

Another method is simply to use a different, but still compatible terminal definition

that is similar to what i am working on.

i often use tmux so my TERM is usually "screen." but it does not work like we want with newterm().

but i see that vt100 has capability smam (automatic margin mode) and screen does not have that capability, for example.

so if i can find the single or handful of capabilities all the working term types have in common we could just, at runtime, see what TERM is bound to and if the user's current TERM has the desirable capabilities then we use newterm() otherwise we just use initscr() and they don't get this "persist the output" feature.

i am looking into that because i am not sure how reliable it is just to change TERM.

it feels more reliable to just say "oh, your current TERM has the right capability so you get the feature otherwise you just get your screen cleared on exit."

[tenox7]

Try this:

infocmp -I $TERM | sed -e 's/smcup=[^,]*,//g' -e 's/rmcup=[^,]*,//g' | tic -

:)

[tenox7]

I have done even more testing and confirmed work around either by specifying different terminal type or patching terminfo works just fine. Changing initscr() to newterm() has nothing to do it. It was probably my poor understanding at that time. I don't think so there is anything that could be done in the code to fix it. Potentially you could implement the fix used by dialog(1) command. If you want send another PR with that. However there is a license incompatibility so please do not copy and paste.