Re: attroff and COLOR_PAIR

[Thomas Dickey]

On Tue, 26 Oct 2010, Dulek wrote:

> Look at this piece of code:
>
> attron(COLOR_PAIR(1));
> addstr("a");
> attron(COLOR_PAIR(2));
> addstr("b");
> attroff(COLOR_PAIR(2));
> addstr("c");
> attroff(COLOR_PAIR(1));
> refresh();
>
> According to  http://www.mkssoftware.com/docs/man3/curs_attr.3.asp manual :

fwiw, the mks page is an old port of ncurses (from the late 1990's iirc).

There's a current set of manpages on my website, e.g.,

http://invisible-island.net/ncurses/man/curs_attr.3x.html

> The routine attroff() turns off the named attributes without turning any
> other attributes on or off.
>
> This means, that this code should display 'a' in COLOR_PAIR(1) color, 'b' in
> COLOR_PAIR(2) color, and 'c' in COLOR_PAIR(1) color. Unfortunately 'c' is
> displayed in default terminal color. Why this is happening?

As far as attron/attroff are concerned, the whole field masked by A_COLOR
is a single attribute.

By the way, this is an extension from X/Open, because it accepts color as 
one of the attributes (matching the SVr4 implementation).  X/Open only 
defines attr_on/attr_off, which separates color from the other attributes 
(because it's allowing for color pairs past the 6-8 bits used for 
implementing A_COLOR).

--
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net