Originally, the compiler complained that the variable was initialized but not used. I set it to zero just to see if it would help, but it merely changed from initialized but not used to set but not used.
players.c:975:86: error: variable 'i' set but not used [-Werror,-Wunused-but-set-variable]
975 | int num = 0, num2 = 0, num3 = 0, num4 = 0, num5 = 0, num6 = 0, num7 = 0, num8 = 0, i = {0}, n_vars = {0};
function load_affects():
Code:
static void load_affects(FILE *fl, struct char_data *ch)
{
int num = 0, num2 = 0, num3 = 0, num4 = 0, num5 = 0, num6 = 0, num7 = 0, num8 = 0, i = {0}, n_vars = {0};
char line[MAX_INPUT_LENGTH + 1];
struct affected_type af = {0};
do {
new_affect(&af);
/* As in load_skills: without this the loop never ends, and each turn
* of it calls affect_to_char(), so the list grows until memory runs
* out rather than merely hanging. */
if (!get_line(fl, line, sizeof(line)))
break;
n_vars = sscanf(line, "%d %d %d %d %d %d %d %d", &num, &num2, &num3, &num4, &num5, &num6, &num7, &num8);
if (n_vars < 1)
break;
/* An affect's spell number runs to TOP_SPELL_DEFINE, which is the size
* of spell_info[] -- not to MAX_SKILLS, which bounds the skills array
* and is a smaller, different thing. SPELL_DG_AFFECT is 298, so a
* player carrying a script-applied affect saves one above MAX_SKILLS
* as a matter of course. */
if (num > TOP_SPELL_DEFINE) {
log("SYSERR: load_affects: spell %d out of range in %s's pfile", num,
GET_NAME(ch) ? GET_NAME(ch) : "a player");
continue;
}
if (num > 0) {
af.spell = num;
af.duration = num2;
af.modifier = num3;
af.location = num4;
if (n_vars == 8) { /* New 128-bit version */
af.bitvector[0] = num5;
af.bitvector[1] = num6;
af.bitvector[2] = num7;
af.bitvector[3] = num8;
} else if (n_vars == 5) { /* Old 32-bit conversion version */
if (num5 > 0 && num5 < NUM_AFF_FLAGS) /* Ignore invalid values */
SET_BIT_AR(af.bitvector, num5);
} else {
/* Not a shape this format has. sscanf leaves the fields it could
* not fill alone, so duration, modifier and location would be
* whatever the record before this one put there -- an affect
* wearing the previous affect's values. It was logged and then
* applied regardless. build_player_index() above refuses a short
* record for exactly this reason. */
log("SYSERR: Invalid affects in pfile (%s), expecting 5 or 8 values", GET_NAME(ch));
continue;
}
affect_to_char(ch, &af);
i++;
}
} while (num != 0);
}
It looks like i is expected to be used, but I think the compiler is optimizing it away. I was able to force the compiler to use i by declaring it a volatile integer.
Code:
static void load_affects(FILE *fl, struct char_data *ch)
{
int num = 0, num2 = 0, num3 = 0, num4 = 0, num5 = 0, num6 = 0, num7 = 0, num8 = 0, n_vars = {0};
volatile int i = {0};
Is i still necessary for this function to work correctly? If it isn't, removing it would be cleanest but it looks like it might be important.
