OpenBSD support

[mischief]

partially fixes #2088 by adding OpenBSD support in the notcurses tree.

while some of the demo programs work, notcurses-demo does not and exits immediately:

fugu$ ./notcurses-demo
0 failed renders, 0 failed rasters, 0 refreshes, 0 input errors
RGB emits:elides: def 0:0 fg 0:0 bg 0:0
Cell emits:elides: 0:0 (0.00%) 0.00% 0.00% 0.00%
Bmap emits:elides: 0:0 (0.00%) 0B (0.00%) SuM: 0 (0.00%)

i'm not sure how to identify what is wrong with it.

[dankamongmen]

exciting! let me take a look. i'll also install an openbsd vm so i can test and assist with debugging. thanks for this work!

[dankamongmen]

i think (and i hope you'll agree) that notcurses-demo is a necessary litmus for considering this a successful extension to openbsd?

the first thing you might want to do is run notcurses-demo -l8 to turn up the logging.

[mischief]

i think (and i hope you'll agree) that notcurses-demo is a necessary litmus for considering this a successful extension to openbsd?

the first thing you might want to do is run notcurses-demo -l8 to turn up the logging.

here's with logging. it does not reveal much more. (level 8 was not valid)

fugu$ ./notcurses-demo -l7
get_tty_fd:482:returning TTY fd 3
notcurses_stop_minimal:176:restored terminal, returning 0
notcurses_drop_planes:1438:we have some planes
ncpile_drop:1419:killing plane 0x9df725b1ee0, next is 0x0
notcurses_drop_planes:1446:all planes dropped
stop_inputlayer:2661:tearing down input thread
free_terminfo_cache:199:brought down input layer
free_terminfo_cache:215:destroyed terminfo cache
0 failed renders, 0 failed rasters, 0 refreshes, 0 input errors
RGB emits:elides: def 0:0 fg 0:0 bg 0:0
Cell emits:elides: 0:0 (0.00%) 0.00% 0.00% 0.00%
Bmap emits:elides: 0:0 (0.00%) 0B (0.00%) SuM: 0 (0.00%)

[dankamongmen]

i think the error output is possibly being hidden. can you try the following two things:

1. run with -k to disable the alternate screen
2. ensure you're supplying a valid data directory with -p (from a build directory without having installed, try ./notcurses-demo -p ../data

and see if anything interesting happens? i do not see myself getting an openbsd vm installed today.

[mischief]

i think the error output is possibly being hidden. can you try the following two things:

1. run with `-k` to disable the alternate screen

2. ensure you're supplying a valid data directory with `-p` (from a build directory without having installed, try `./notcurses-demo -p ../data`

and see if anything interesting happens? i do not see myself getting an openbsd vm installed today.

using ./notcurses-demo -p ../data -k -l7 2> notcurses-log.txt, here is the log:
notcurses-log.txt

near the end is

handoff_initial_responses_late:1209:handing off initial responses
block_on_input:2490:blocking on input availability
block_on_input:2545:waiting on 1 fds (ibuf: 0/1024)
apply_term_heuristics:986:tname is screen (qterm 8)
ncplane_new_internal:682:created new 61x213 plane "std" @ 0x0
notcurses_drop_planes:1438:we have some planes
ncpile_drop:1419:killing plane 0xcc74e142e00, next is 0x0
notcurses_drop_planes:1446:all planes dropped
utf8_egc_len:139:prohibited or invalid unicode: 0x00000000
notcurses_stop:1450:stopping notcurses
notcurses_stop_minimal:176:restored terminal, returning 0
notcurses_drop_planes:1438:we have some planes
ncpile_drop:1419:killing plane 0xcc74e142e00, next is 0x0
notcurses_drop_planes:1446:all planes dropped
stop_inputlayer:2661:tearing down input thread
free_terminfo_cache:199:brought down input layer
free_terminfo_cache:215:destroyed terminfo cache
0 failed renders, 0 failed rasters, 0 refreshes, 0 input errors
RGB emits:elides: def 0:0 fg 0:0 bg 0:0
Cell emits:elides: 0:0 (0.00%) 0.00% 0.00% 0.00%
Bmap emits:elides: 0:0 (0.00%) 0B (0.00%) SuM: 0 (0.00%)

is utf8_egc_len:139:prohibited or invalid unicode: 0x00000000 making the demo terminate?

[dankamongmen]

is utf8_egc_len:139:prohibited or invalid unicode: 0x00000000 making the demo terminate?

nope (it's coming after notcurses_drop_planes(), which is in the shutdown sequence. it's unexpected, though. perhaps that's coming from the input thread?

i see you're running in screen. does this happen outside of screen?

[mischief]

is utf8_egc_len:139:prohibited or invalid unicode: 0x00000000 making the demo terminate?

nope (it's coming after notcurses_drop_planes(), which is in the shutdown sequence. it's unexpected, though. perhaps that's coming from the input thread?

i see you're running in screen. does this happen outside of screen?

it's OpenBSD tmux (sets TERM=screen, same behavior with screen-256color). outside of tmux the behavior is identical - linux TERM=st-256color -> ssh to OpenBSD 7.8: it starts and then exits without drawing.

[mischief]

checked xterm (TERM=xterm-256color), same behavior.

[dankamongmen]

thanks. i'll need to set up an openbsd vm to debug this. i appreciate you running those tests and starting this work!

[mischief]

is utf8_egc_len:139:prohibited or invalid unicode: 0x00000000 making the demo terminate?

so the bug is here.

gdb$ bt
#0  _libc_mbrtowc (pwc=0x71b5e92a36ec L"__\xaedc427_\xaedebf8_\x3075810_\x3075179_\xaedebee_\x7cea4d84\x57af24aa\xe92a3750__\xf0cb31b_\xaedebee0", s=0xe343075179b "", n=0x4, ps=<optimized out>) at /usr/src/lib/libc/locale/multibyte_citrus.c:53
#1  0x00000e36f0cb4054 in utf8_egc_len (gcluster=0xe343075179b "", colcount=0x71b5e92a3734) at /home/mischief/src/notcurses/src/lib/egcpool.h:113
#2  0x00000e36f0cb31b6 in nccell_load (n=0xe36aedebee0, c=0xe36aedebf80, gcluster=0xe343075179b "") at /home/mischief/src/notcurses/src/lib/notcurses.c:1838
#3  nccell_prime (n=0xe36aedebee0, c=0xe36aedebf80, gcluster=0xe343075179b "", stylemask=0x4, channels=0x71b5e92a3650) at /home/mischief/src/notcurses/include/notcurses/notcurses.h:756
#4  ncplane_set_base (ncp=0xe36aedebee0, egc=0xe343075179b "", stylemask=0x4, channels=0x71b5e92a3650) at /home/mischief/src/notcurses/src/lib/notcurses.c:1569
#5  0x00000e34307a4424 in scrub_stdplane (nc=0xe36aedc4270) at /home/mischief/src/notcurses/src/demo/demo.c:507
#6  main (argc=<optimized out>, argv=<optimized out>) at /home/mischief/src/notcurses/src/demo/demo.c:568

quoth the manual

RETURN VALUES
     0             The bytes pointed to by s form a terminating NUL character.
                   If wc is not NULL, a NUL wide character has been stored in
                   the wchar_t object pointed to by wc.

we pass an empty string so, in utf8_egc_len mbrtowc returns 0, which is not handled. then we fall through to

        if(iswcntrl(wc)){                                                                                                                                     
          logerror("prohibited or invalid unicode: 0x%08x", (unsigned)wc);                                                                                    
          return -1;                                                                                                                                          
        }                                                                                                                                                     

where iswcntrl returns true for the null byte.

i admit i am wholly unfamiliar with wide/multibyte character programming. should utf8_egc_len break the loop when mbrtowc returns 0?

[mischief]

i added my change to handle NUL bytes from mbrtowc, and dropped the change to CMakeLists.txt since commit openbsd/src@80c5435 upstream added ncursesw pkg-config files.

@dankamongmen can you take another look?