Game: OpenD6 FUNC_2 adds the pips of both attributes instead of averaging them #109

FUNC_2 in globalScripts/Convert_OPEND6_FANTASY.gd averages two OpenD6 attributes (dice codes like 3D+1) and converts the average to a backend percentile. It's used for CH (CHARISMA/ACUMEN) and SD (ACUMEN/PHYSIQUE).

It halves the dice (die = dieCombo/2), but the pips (the +N) from both scores are added together and never halved. The comment says "First average dice (add/2; remainder to +; ...)", but when the dice add up to an odd number, the leftover half die is dropped instead of going into the pips.

As a result, the "average" of two equal scores isn't equal to either one, and can land on either side of them:

For reference, FUNC_1 converts nD+p to (n - 1) × 25 + p × 10.

Possible fixes

A. Average in pips, rounding down. Turn each code into pips (3 × dice + pips, since 3 pips = 1D in OpenD6), average them, round down, convert back to nD+p, then apply the FUNC_1 formula.

var pips = (3 * die1 + extra1 + 3 * die2 + extra2) / 2   # integer division
var die = pips / 3
var extra = pips % 3

B. Average in pips, rounding half up. Same as A, but a leftover half pip rounds up: (total_pips + 1) / 2.

C. Average the two percentiles. Convert each score with FUNC_1, then average the results:

return int(round((FUNC_1(score1) + FUNC_1(score2)) / 2.0))

All three fix the main bug: two equal scores now give the same result as FUNC_1.

Suggestion: option C

• It's the most accurate. It's the exact average of what each attribute converts to, off by at most 0.5 from rounding. A and B round to whole pips, and one pip is worth 10 percentile points, so they can be off by 5 or more. In the 3D / 2D row, the true average is 37.5, and A and B give 35 and 45.
• It avoids a quirk in FUNC_1. One pip is worth 10 points, but one die is only worth 25, not 30. So 2D+2 (45) and 3D (50) are only 5 points apart. Averaging in pips and then converting carries that unevenness into the result. Averaging after conversion doesn't.
• It's consistent. D20_Converter.gd already averages percentiles, and Game: BFRPG FUNC_2 drops the half before rounding when averaging two abilities #108 suggests the same approach for BFRPG. It also removes the hand-rolled dice and pip arithmetic, including the "+3 and up adds a die" carry.
• The backend doesn't need values on the OpenD6 grid. Backend capabilities are 0–100 percentiles checked with a d100 roll-under, so a value like 38 works fine.

If results should stay on values a real OpenD6 code could produce, A is the OpenD6-native choice. It matches how OpenD6 treats dice and pips, and it's closest to what the existing comment describes.

@MacDugRPG, you wrote this converter. Does C work for you, or do you want results that map back to real OpenD6 codes (A or B)? Once that's settled, I'm happy to fix it.