Re: ncurses mvaddch() cursor movement a possible bug?

[Tim van der Molen]

Avinash Sonawane (2015-03-30 12:41 +0200):
> Hello there! I have a query regarding mvaddch cursor movement. Please
> go through the following code snippet. I am using xterm BTW.
> 
> Code:
> 
> #include <ncurses.h>
> 
> int main(int argc, char **argv)
> {
>     initscr();
>     noecho();
>     cbreak();
> 
>     mvprintw(0, 0, curses_version());
>     mvprintw(1, 0, "Hello World");
>     mvaddch(2, 0, mvinch(1, 4));                 // Why doesn't this work?
> 
>     getch();
>     endwin();
> 
>     return 0;
> }
> 
> Output:
> 
> ncurses 5.9.20130608
> Hello World
> 
> with pointer blinking (waiting for getch ) just after o of Hello.
> 
> Question:
> As in C, arguments passed to a function are evaluated first before
> calling that function, mvinch() will be called first and when it'll
> return the character o the call to mvaddch() will be made.
> But then why character o is not printed on line 2 (just below Hello
> World)? Instead mvaddch prints o at current cursor position (thanks to
> winch which is 1,4). Here mvaddch() behaves just like addch paying no
> respect to the mv prefix and the explicit movement coordinates given
> to it. Why?
> 
> Is this a possible bug in mvaddch() or am I missing something?

mvaddch() and mvinch() are not functions, but macros. If you run your
code through cpp, you'll see that mvaddch(2, 0, mvinch(1, 4)) expands
to:

        (wmove(stdscr,2,0) == (-1) ? (-1) :
        waddch(stdscr,(wmove(stdscr,1,4) == (-1) ? (chtype)((-1)) :
        winch(stdscr))));