Rational errors fix

[pk-2000]

• Better debug information on float errors (lines 231 and 251).
• line 246 fixes ruby rational errors

[cremno]

Have you tested this? FIX2INT() only works with fixnums.

[pk-2000]

Yes,
the only games that I encountered this problem are some rgss1 games,
that use a plugin to determine the characters position.
float value (X,Y position of the character) & rational value (expected X,Y new value)
The new value is determined from the move direction of the character.

[cremno]

Yeah, RGSS3 also supports rational numbers but your change may only appear to work. Could you please try something like Rect.new(Rational(10, 10), 0, 0, 0) and check the value of its x attribute? I can't compile mkxp at the moment and there's no test suite or build bot.

IMHO the type check isn't necessary anyway. *out = NUM2INT(arg); is good enough (and rb_rescue() if the custom error message should be kept). This would add support for complex numbers too: Rect.new(Complex(10, 0), 0, 0, 0).

[pk-2000]

Can you give another test?
My Falcon-mkxp has unlocked resolution, so the x result will always deffer, depending on the game resolution & monitor resolution & position of the window.

• I had tested it against these examples
  https://ruby-doc.org/core-2.5.3/Rational.html
  and the only ones that could not resolve (if I remember correctly) were the examples with half keyword arguments e.g.
  Rational(25, 100).round(1, half: :up) #=> (3/10)

[cremno]

X should always be 1. But it can be tested with any other integer attribute:

s = Sprite.new
s.z = Rational(10, 2)
p s.z # => 5

[pk-2000]

Thank you.
I created 2 events to call the test scripts and both of them
return positive random(?) values with 8 digits(!) instead of returning x=1 & s=5.