[r2038] Weird randart thing

[miyazaki]

Found this randart in the spoilers:

The Spear 'Rhorlata' (3d6) (-11,-7)
-----------------------------------
Heavily cursed.
Slays giants.
Cannot be harmed by acid, electricity, fire, cold.
Aggravates creatures nearby.  Drains experience.  Induces random
teleportation.  


Min Level 1, Max Level 0, Generation chance 40, Power 4294967291, 5.0 lbs

[juggle5]

Looks like a signed/unsigned int bug. If the power was set to -5 (is a negative power allowed?), but printed as an unsigned integer, then the output would be 4294967291.

[emulord]

Looks like the Power calculation overflowed to negative. Thats near MaxUnsignedLong. It wont ever generate, but if it could it would be worth a lot according to power based pricing (??)

[Pete Mack]

No overflow, just a formating problem in the log file.
There are a handful of negative power artifacts. (Nobody sane would use Mormegil or Beruthiel, unless it's some kind of crazy challenge.)

[miyazaki]

I wondered more about the max depth.

[Magnate]

I wondered more about the max depth.

Max depth is also based on power, so anything with negative power will have max depth 0. I will fix this soon, so max depth is never below 100.

@Pete: what's the right %foo formatting to print a signed int instead of unsigned?

[Pete Mack]

@magnate:
In the unlikely event that you haven't figured this out already:

printf("%u", i);

[Magnate]

@magnate:
In the unlikely event that you haven't figured this out already:

printf("%u", i);

Thanks Pete - how fabulously counter-intuitive is that! %d for UNsigned, %u for signed!

[PowerDiver]

Thanks Pete - how fabulously counter-intuitive is that! %d for UNsigned, %u for signed!

Why would you ever want to do that? The only time I care how I print unsigned stuff is bit flags or pointers or the like, and those I want in hex. The standard %d works fine unless the high order bit is 1. If you have real unsigned stuff with high bit of 1, you are playing with fire because you'll have overflow errors sooner or really sooner.

If you have an unsigned char packed into a normal char, %d combined with a cast of the argument works fine.

[d_m]

Thanks Pete - how fabulously counter-intuitive is that! %d for UNsigned, %u for signed!

I think you have this backwards:

#import <stdio.h>
int main(void) {
    unsigned int i = -1;
    printf("%d %u\n", i, i);
    return 0;
}

Produces:

-1 4294967295

[Pete Mack]

Sorry Magnate, I misread. %u is Unsigned decimal, %d is signed Decimal.

[Magnate]

Sorry Magnate, I misread. %u is Unsigned decimal, %d is signed Decimal.

Then something odd is going on. This all started because negative power ratings get printed out unsigned in the artifact spoiler ... but I didn't know about %u, so I'm certain I'm using %d ...

[Pete Mack]

Did you verify that you are printing a signed number in the first place?

If you are casting--say--(byte) to (int), you will always get an unsigned integer.
You may need to cast to (char) first: (int)((char)x)

Edit:
Fixed, per Zaimoni's comment.

[zaimoni]

Did you verify that you are printing a signed number in the first place?

If you are casting--say--(byte) to (int), you will always get a signed integer.
You may need to cast to (char) first: (int)((char)x)

Actually, on normal platforms casting byte/unsigned char to int will be a strictly positive number after the cast.

Casting signed char to int works as expected. Of course, since it is target-dependent whether char acts like signed char or unsigned char things can go weird.

[Pete Mack]

@Zaimoni--
My bad: I meant to say: you will always get the signedness of the original value.
Fixed in original.