Integer division breaks Magic Finder

[1da1a172]

AndorsTrail/src/com/gpl/rpg/AndorsTrail/controller/SkillController.java

private static int getRollBias(ConstRange chance, Player player, SkillID skill, int perSkillpointIncrease) {
        int skillLevel = player.getSkillLevel(skill);
        if (skillLevel <= 0) return 0;
        return chance.current * skillLevel * perSkillpointIncrease / 100;
}

Here for magic finder, skillLevel is the level of the Magic Finder skill, perSkillpointIncrease is 50, and chance.current is (usually?) 1, from the item chance. Notably, these are all ints, as is the return type. So if we have a skillLevel of 1, we get: 1 * 1 * 50 / 100 = 0 because integer division truncates. Even if these were floats or doubles, it wouldn't matter, though.

AndorsTrail/src/com/gpl/rpg/AndorsTrail/controller/Constants.java

public static boolean rollResult(final ConstRange r, int bias) { return rollResult(r.max, r.current + bias); }
...
private static boolean rollResult(final int probabilityMax, final int probabilityValue) { return rnd.nextInt(probabilityMax) < probabilityValue; }

The random number is always an integer, so even if probabilityValue were 1.5f, that extra 0.5 isn't ever going to help.

Suggestions: 1) Change the description of magic finder to indicate that you are only getting a bonus for every 2 skill points 2) Don't use ints for the calculations (make sure you get them all!) 3) In src/com/gpl/rpg/AndorsTrail/model/ability/SkillCollection.java, change PER_SKILLPOINT_INCREASE_MAGICFINDER_CHANCE_PERCENT to 100 (which ends up being a roundabout way of just setting the bias to the skill level)

[Chriz76]

Thanks for reporting this. You can also fix it yourself if you want and be part of the community?

[1da1a172]

I wasn't sure which direction the devs wanted to go with it. As a player, I obviously want option 3, but the devs probably have a better idea of the balance of the game.

[Chriz76]

Good point. We will check and discuss it

[1da1a172]

The random number is always an integer, so even if probabilityValue were 1.5f, that extra 0.5 isn't ever going to help.

I just realized that isn't entirely true. the extra 0.5 will always help, since it is a <, not a <=. It just isn't having the effect one might think.

[Chriz76]

It might be possible to change the chance e.g. from 1.5/100 to 3/200

[Rijackson]

Not option 1 for sure! That will really annoy players that took MF. Number 3 seems the easiest, as long as here are no consequences elsewhere.

[Chriz76]

I would prefer if you find a way for option 2.

[Chriz76]

There are only three calls using the bias. So it might be easy to change int getXXXBias(...) to ConstRange getModifiedXXXChance(chance, ...)

[Chriz76]

Should be fixed and will be part of the next release: https://github.com/NutAndor/andors-trail/commit/6dc33bcac2907b806b1aa0d24569231126910ebb