Hmmm, I think it may be as you said.. this is what I have at the end of fread_char() in save.c:

for (paf = ch->affected; paf != NULL; paf = paf->next)
    {
        if (paf->type < 0 || paf->type >= MAX_SKILL)
            continue;

        fprintf (fp, "Affc '%s' %3d %3d %3d %3d %3d %10d\n",
                 skill_table[paf->type].name,
                 paf->where,
                 paf->level,
                 paf->duration, paf->modifier, paf->location, paf->bitvector);
    }

    fprintf (fp, "End\n\n");
    return;
}

What should it look like? Just duplicate and make it "ch->affected2" and "Affc2" ?

No, you don't want to make a 2nd affected list. Really, this part of the code should not have to
be changed (and the code here for writing does look right) for the affected2_by changes. I would
do a couple things, first, I would check the pfile and make sure it is actually saving there, then
I would have another spell affect (so you have at least two, and make the other something other
then an affected2_by spell effect) when you log off to see if it is a problem with all spell effects,
or just the affected2_by ones. Another thing I would check (though is a long shot) is to make sure
gsn_focus is valid (in the range of 0 to MAX_SKILL).

Upon further investigation… it appears that the affect IS saved when logging on/off..

but it is when the spell fades that it doesn't completely disappear..

Here's what I mean:


I also checked in the pfile before and after the affect wore-off..

Before:

Gr 'weaponsmaster'
Affc 'focus'   6  27  19   0   0          2
End

After:

Gr 'weaponsmaster'
End

I thought this might help… no idea what is going on with this stupid thing:

interp.h"(140,20):	DECLARE_DO_FUN(	do_focus		); 
merc.h"(1012,13):	#define AFF_FOCUS			(B)
merc.h"(2100,20):	extern sh_int  gsn_focus;
act_info.c"(3119,9):	void do_focus (CHAR_DATA * ch, char *argument)
act_info.c"(3123,31):   if (IS_AFFECTED2 (ch, AFF_FOCUS))
act_info.c"(3125,34):	send_to_char ("You are already focusing on your surroundings.\r\n", ch);
act_info.c"(3129,32):	send_to_char ("You attempt to focus on your surroundings.\r\n", ch);
act_info.c"(3131,48):   if (number_percent () < get_skill (ch, gsn_focus))
act_info.c"(3133,29):	check_improve (ch, gsn_focus, TRUE, 3);
act_info.c"(3135,23):   af.type = gsn_focus;
act_info.c"(3140,28):   af.bitvector = AFF_FOCUS;
act_info.c"(3145,32):   check_improve (ch, gsn_focus, FALSE, 3);
act_info.c"(3146,29):	send_to_char ("You try to focus but get distracted.\r\n", ch);
act_move.c"(280,63):	|| (IS_AFFECTED2 (ch, AFF_CREEP) && IS_AFFECTED2 (fch, AFF_FOCUS))))
act_move.c"(299,63):	|| (IS_AFFECTED2 (ch, AFF_CREEP) && IS_AFFECTED2 (fch, AFF_FOCUS))))
const.c"(2051,7):     	"focus", {53, 53, 1, 12}, {0, 0, 4, 6},
const.c"(2053,11):     	&gsn_focus, SLOT (0), 0, 12,
const.c"(2054,29):     	"", "You begin to lose focus on your surroundings.", ""},
db.c"(161,12):		sh_int gsn_focus;
handler.c"(2713,30):	&& (!IS_AFFECTED2 (ch, AFF_FOCUS)
handler.c"(2903,19):	if (vector & AFF_FOCUS)
handler.c"(2904,24):    strcat (buf, " focus");
interp.c"(271,7):    	{"focus",	do_focus, POS_STANDING, 0, LOG_NORMAL, 1},
tables.c"(170,4):	{"focus", B, TRUE},

Ok, that would point more to the part of affect_modify() (handler.c) that is supposed to remove
the affected2_by effect flags. Close to the top of affect_modify() there should be an if based
on fAdd, the first (true) part of the if is for when you are adding an effect, and the else (false)
part is for when you are removing an effect. That false part should have a case for TO_AFFECTS2
that does a REMOVE_BIT of the ch->affected2_by flag set.

As a note here. You would also get the same thing if the character was also wearing an object
that gave the affected2_by flag (maybe put on after the spell was cast). I don't see any evidence
here that is the case, but thought I should mention the possibility.