386WGS console driver bug?

[Flint Pellett]

/*
** I believe that I "may" have found a bug in the AT&T console driver,
** or something related. 
**
** Here is my sample curses program that demonstrates the problem.
** WARNING WARNING: executing this program will leave your terminal
** in the alternate charset when you are returned to the shell: even tho
** it will all look like gibberish, you ought to be able to get back to
** normal with a "tput rmacs" or a "tput init", but I can't make any
** guarantees about your system: be prepared to reboot.  This behavior is
** exhibited on an AT&T 6386 WGS running AT&T UNIX System V R3.2.2.
**
** The program merely tries to draw two vertical lines 6 columns apart.
** The problem is, because both the A_ALTCHARSET bit and the A_BOLD bit
** are on to plot the ACS_VLINE chars, it doesn't get reset back out of
** altcharset mode after plotting the first VLINE char: so it plots
** 0xA0 (An a with an accent quote above it) between the vertical lines
** when it echoes spaces to position to the 2nd column, and then when it
** exits the program it leaves you in the alternate charset.
*/
#include <curses.h>

main()
{
	initscr();

	attrset(A_BOLD);	/* The KEY to the problem */

	mvaddch( 11, 10, ACS_VLINE );	/* row 11, col 10 */
	mvaddch( 11, 15, ACS_VLINE );	/* row 11, col 15 */

	refresh();

	endwin();
}

/***
Here is the od -c of the trapped output to show you what it is doing:

0000000 033   [   0   m 033   [   2   J 033   [   H 033   [   1   2   ;
0000020   1   1   H 033   [   1   2   m 033   [   1   m   3 033   [   0
0000040   m                 033   [   1   2   m 033   [   1   m   3 033
0000060   [   0   m 033   [   2   5   ;   0   1   H 033   [   0   m  \0
0000077

You can see on line 2 that it echoed an \E[12m for alternate chars,
then \E[1m (bold) then the "3" for the vertical line.  It then sent
the sgr0 string \E[0m which is supposed to mean "reset all attributes"
but which only reset the bold attribute.  When I changed my terminfo
to have sgr0=\E[0m\E[10m  (ie, I added an explicit rmacs to the sgr0)
it worked properly.  So: if \E[0m is supposed to include "alternate
charset" as one of "all attributes", then there is a bug in it.  I
assume that \E[0m is supposed to do that, because if you do not have
the bold attribute selected, then \E[0 _will_ turn off the alternate
chars attribute properly: it is only when bold is also on that it
doesn't get the job done.
***/
-- 
Flint Pellett, Global Information Systems Technology, Inc.
1800 Woodfield Drive, Savoy, IL  61874     (217) 352-1165
uunet!gistdev!flint or flint at gistdev.gist.com