Truncation

ID:1476314
 
Jan 20 2014, 3:43 pm
Truncates a number's decimal component. Useful for converting numbers to integers.

proc/trunc(x)
  return x && x / abs(x) * round(abs(x))


I wrote this because I saw somewhere on the forums that someone recommended round()(whose default format is floor()) to truncate a number. That's incorrect. The reasoning for this is simple: truncating rounds towards zero and round() rounds towards negative infinity.

round(-3.14159) // -4
trunc(-3.14159) // -3

// The only exception!
round(3.14159) // 3
trunc(3.14159) // 3




Jan 20 2014, 4:02 pm
round(num,1) will round towards 0 in the case of negatives.

round(-3.14159,1) // -3
Jan 20 2014, 4:37 pm
Oh, wow. If you think about it, that does make sense. Ah well, now people will have two options to choose from, with the latter being faster. :(
Jan 20 2014, 4:51 pm
In response to Nadrew
Nadrew wrote:
round(num,1) will round towards 0 in the case of negatives.

> round(-3.14159,1) // -3
>


False

round(-3.8, 1) // -4
Jan 20 2014, 5:02 pm
In response to Lugia319
It's only partially false. It'll round up if the first digit after the decimal point is >5 and it'll round down(towards zero) when <5.

Hence why 3.14149 became 3 when round()ed.
Jan 20 2014, 5:24 pm
Partially false is like partially pregnant.

round(x, 1) is pretty much the rounding we're used to. Rounding towards the nearest whole number.
Jan 20 2014, 5:48 pm (Edited on Jan 22 2014, 9:48 am)
Lugia's right. Rounding towards zero is different from rounding to the nearest whole number, floor, and ceil.

Rounding towards zero would be sign(n) * floor(abs(n)), which is what Magnum showed.