Windows v6 beta test

[fowlplay4]

I'm currently having issues with a scripted timevar3 system, and have narrowed it down to the math involving large numbers to be at fault.

PHP Code:

//#CLIENTSIDE

function onCreated() {
  for (temp.i = 1; temp.i < 10; temp.i++) {
    this.scheduleevent(1 * temp.i, "MathBug", "");
  }
}

function onMathBug() {
  temp.largeNumber = 1274191251.03; // example server-side timevar2
  echo("=========================");
  echo(temp.largeNumber);
  echo(timevar2);
  echo(temp.largeNumber+timevar2);
} 


Log:

PHP Code:

=========================
1274191251.029999972
186.784149184
1274191488
Should of been: 1274191437.814149156
=========================
1274191251.029999972
187.783325184
1274191488
=========================
1274191251.029999972
188.781326272
1274191488
=========================
1274191251.029999972
189.77952576
1274191488
=========================
1274191251.029999972
190.77836608
1274191488
=========================
1274191251.029999972
191.776428224
1274191488
=========================
1274191251.029999972
192.774490368
1274191488
=========================
1274191251.029999972
193.773513792
1274191488
=========================
1274191251.029999972
194.78767392
1274191488 


[Tigairius]

Originally Posted by fowlplay4

I'm currently having issues with a scripted timevar3 system, and have narrowed it down to the math involving large numbers to be at fault.

This is not really a bug though. The maximum integer value that Graal supports is 2^32, which is 1274191488, due to memory restrictions.

[Twinny]

I had issues doing comparisons of larger numbers with V6 client. Skyld showed me that comparing them as strings fixes the issue.

[Tigairius]

Originally Posted by Twinny

I had issues doing comparisons of larger numbers with V6 client. Skyld showed me that comparing them as strings fixes the issue.

Yes, the limit for strings is much larger than for integers.

[WhiteDragon]

Maybe some arbitrary-precision arithmetic is in order? I'm not sure how this would be handled due to GS2's dynamic typing though -- it could be a very hard-to-find bug if your script suddenly slows down once the numbers go above 2^32.