Re: [AD] num lock dependent scancode conversion

[Peter Wang <tjaden@xxxxxxxxxx>]

stephen.kittelson@xxxxxxxxxx <stephen.kittelson@xxxxxxxxxx> wrote:
> 
> Is my suggestion for scancode conversion, that has to be enabled by
> setting a flag, rejected?

I don't plan to speak for those who hold the keys to CVS, but here
are my thoughts and a solution.

1. It is not a commonly requested feature and although it is only
enabled through a flag, it still bloats the library,

2. The patch was a hack (no offense, but `keypad_keysdn' was like
fitting a square peg through a round hole),

3. It can be achieved just as easily and comfortably in userspace
with something like the attached (which affects `ureadkey').

-- 
tjaden@xxxxxxxxxx - http://www.psynet.net/tjaden/
BRIMPLET (BRIM plit), n.  A frayed shoelace that must be moistened to pass
through a shoe eyelet.  -- Rich Hall, "Sniglets"

#include <allegro.h>

static int ucallback (int key, int *sc)
{
    if (!(key_shifts & KB_NUMLOCK_FLAG)) 
	switch (*sc) {
	    case KEY_0_PAD: *sc = KEY_INSERT; break;
	    case KEY_1_PAD: *sc = KEY_END; break;
	    case KEY_2_PAD: *sc = KEY_DOWN; break;
	    case KEY_3_PAD: *sc = KEY_PGDN; break;
	    case KEY_4_PAD: *sc = KEY_LEFT; break;
	    case KEY_6_PAD: *sc = KEY_RIGHT; break;
	    case KEY_7_PAD: *sc = KEY_HOME; break;
	    case KEY_8_PAD: *sc = KEY_UP; break;
	    case KEY_9_PAD: *sc = KEY_PGUP; break;
	}

    return key;
}

END_OF_STATIC_FUNCTION (ucallback);

int install_nld_ureadkey ()
{
    if (!keyboard_driver)
	return -1;

    LOCK_FUNCTION (ucallback);
    keyboard_ucallback = ucallback;
    return 0;
}