[PATCH 14/45] man/ncurses.3x: Change application template to use `noqiflush()` instead of `intrflush()`.

[G. Branden Robinson]

This is a potentially controversial change.

1.  `intrflush()` had a bad interface; it demanded an argument that it
    did not use.
2.  More debatable, maybe, is the recommendation of `noqiflush()` over
    `qiflush()`.  However the proffered reason for using `qiflush()` is
    performance, and in the 2020s ncurses applications are less often
    used over (slow) serial lines compared to the 1990s.  It seems to me
    that keeping the library's notion of the screen's contents coherent
    with its actual appearance[*] is more valuable as a default case for
    novice users of the library.

[*] Assuming nothing else writes to the terminal device, a problem no
    curses can solve without device locks.
---
 man/ncurses.3x | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/man/ncurses.3x b/man/ncurses.3x
index 7f005e5cc..02d9478e6 100644
--- a/man/ncurses.3x
+++ b/man/ncurses.3x
@@ -213,11 +213,11 @@ .SS "Application Structure"
 .EE
 .RE
 .PP
-Most applications perform further setup as follows.
+Most applications would perform further setup as follows.
 .PP
 .RS 4
 .EX
-intrflush(stdscr, FALSE);
+noqiflush();
 keypad(stdscr, TRUE);
 .EE
 .RE
-- 
2.30.2

signature.asc
Description: PGP signature