"+=" inconsistency.

[JesseParticipant]

when adding a float to an integer with “+=” gives odd results:

Monkey

[crayon-5cba15495b653371797777 inline="true" ] Namespace myapp #Import "<std>" Using std.. Function Main() Local v:Vec2i = New Vec2i v.x += .1 Print v.x Local y:int y += .1 Print y End

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

[crayon-5cba15495b653371797777 inline="true" ] Namespace myapp   #Import "<std>"   Using std..   Function Main()   Local v:Vec2i =  New Vec2i v.x += .1 Print v.x Local y:int y += .1 Print y End

[/crayon]

This should print 0 for both but it displays some odd numbers and inconsistent at that.

It might be an issue for OSX Only.

I already posted it to GitHub. This is just in case somebody runs into this problem as well.

[TomToadParticipant]

I get 11598408 for both on Windows 10.  I think it has to do with the Float to Int conversion done on the .1. If you use 0.1 instead or define y as a float, then you get correct results.

[JesseParticipant]

I get inconsistent results every time I run it:
this:
-1718160848
-1718160848
then this:
-1842821072
-1842821072
then this:
-1145541584
-1145541584

etc..

The reason I run into this problem was because I was trying to get some integer values from a Rnd operation. and I couldn’t figure out why I was getting this strange values. I was getting this by adding the random float value to an integer.

[DaniloParticipant]

It has nothing to do with ‘+=’, it is that ‘.1’ is not correctly converted to integer.
The result is a random (uninitialized) number:

Monkey

Function IntFunc( value:Int ) Print value End Function Main() For Local i:Int = 1 To 5 IntFunc(.1) ' Shows the same value 5 times Next Print "----------" Local x:Int ' Display 5 times different value x = .1 Print x x = .1 Print x x = .1 Print x x = .1 Print x x = .1 Print x End

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Function IntFunc( value:Int )     Print value End   Function Main()       For Local i:Int = 1 To 5         IntFunc(.1) ' Shows the same value 5 times     Next          Print "----------"          Local x:Int     ' Display 5 times different value     x = .1     Print x     x = .1     Print x     x = .1     Print x     x = .1     Print x     x = .1     Print x End

For the loop it displays the same number 5 times, because it is only generated once.
For the Print statements, each of the 5 assignments result in different uninitialzed/random numbers.

Output:

Monkey

17129798 17129798 17129798 17129798 17129798 ---------- -1467958384 -1467957952 -1467957504 -1467957056 -1467956608

1 2 3 4 5 6 7 8 9 10 11 12

17129798 17129798 17129798 17129798 17129798 ---------- -1467958384 -1467957952 -1467957504 -1467957056 -1467956608

The generated C++ code has the wrong numbers directly implemented (hard coded):

C++

{ bbInt l_i=1; for(;(l_i<=5);l_i+=1){ g_default_IntFunc(4312097094); } } bb_print(bbString(L"----------",10)); bbInt l_x{}; l_x=140538451901328; bb_print(bbString(l_x)); l_x=140538451901760; bb_print(bbString(l_x)); l_x=140538451902208; bb_print(bbString(l_x)); l_x=140538451902656; bb_print(bbString(l_x)); l_x=140538451903104; bb_print(bbString(l_x)); } void g_default_IntFunc(bbInt l_value){ bb_print(bbString(l_value)); }

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

{     bbInt l_i=1;     for(;(l_i<=5);l_i+=1){       g_default_IntFunc(4312097094);     }   }   bb_print(bbString(L"----------",10));   bbInt l_x{};   l_x=140538451901328;   bb_print(bbString(l_x));   l_x=140538451901760;   bb_print(bbString(l_x));   l_x=140538451902208;   bb_print(bbString(l_x));   l_x=140538451902656;   bb_print(bbString(l_x));   l_x=140538451903104;   bb_print(bbString(l_x)); }   void g_default_IntFunc(bbInt l_value){   bb_print(bbString(l_value)); }

[JesseParticipant]

I see. But strange still.
I was going to post the exact code but I have since modified it and can’t exactly remember how I had it that it was giving me wrong values with a Rnd assignment to an integer. I tried to duplicate it but can’t do it anymore. It wasn’t a constant value.
Oh well it will probably will creep in some other project again.

[DaniloParticipant]

> But strange still.

Of course still a bug that needs to be looked at by Mark.
Works with ‘0.1’, but with ‘.1’ there is a problem somewhere
in the semantic process, automatic type conversion, or
code generation.

[Author]

Posts