Oxygene: / and div operators

Elements
#1

Hello,
Regarding the documentation (https://docs.elementscompiler.com/Oxygene/Expressions/ArithmeticAndLogicalExpressions/#arithmetic-operators) it is not clear what the difference is (and I think that it must be clarified in the documentation).

What I understand:

  • “/” makes a rounded division
  • “div” makes a truncated division (takes integer part of the division, without rounding).

But, for the rounded division: is it rounded towards zero or rounded towards infinity?
This makes a huge difference for negative numbers: in the first case “-3 / 2” returns -1, in the second one, it returns -2.

Note: doing a test, both (-3 / 2) and (-3 div 2) return -1. So why there is a difference in the description of the two operators in the documentation?

#2

IIRC Both operators behave the same way int regular Oxygene mode, with the operands deciding the result type (eg integer/interg -> integer, integer/double -> double). In Delphi Compatibility Mode, / will always create a double, while div (iirc) always creates an integer — eseetally they behave the same way they do in Delphi.

it should be rounded according to proper math rules. ice 2.4 -> 3,. 2.5 -> 3

If only there were a way to find out :wink:

#3

So, the documentation must not be different between the two operators:

#4

Not outside of Delphi Compatibility no. I’ll review.

#5

Thanks, logged as bugs://84528

#6

So the way this is done:

In Delphi mode:

  • / always does float div; ie 10 / 3 = 3.33333333
  • div always does int divide, and requires integers on left and right side.

in non Delphi mode:

  • / and DIV work the same:
    • If there’s a float on the left or right side, it’s a float
    • Else it’s an integer
1 Like
#7

But my question was about rounding when doing integer division.
And why the documentation is not the same for both operators.

#8

The documentation doesn’t say that: “Divide” only, not a remark for Delphi compatibility mode, as for the “/” operator.

#9

Yes. Just gave summary though. As for the actual rounding, it rounds towards zero. We will look into fixing the docs of course.

#10

That’s what bugs://84528 was logged for, FTR, I just wanted Carlo’s final feedback before deciding whether to doc what it does now vs changing anything compiler side.

#11

bugs://84528 got closed with status nochangereq.