Expressions like 1.23*10^15 give the wrong number (tries to perform 10^15 as an integer that overflows).

[ksp-kos]

Expressions like 1.23*10^15 give the wrong number (tries to perform 10^15 as an integer that overflows).

[Dunbaratu]

A similar problem happens with multiplication of big integers:

set x to 100000000.  // small enough to fit in an int32
print x.
100000000.
set y to 500000000. // also small enough to fit in an int32
print y.
500000000.
print x * y.  // too big for an int32
784662528.

I'm not sure what a good solution to the problem should be. On the one hand programmers have dealt with this for years and know the difference between integers. long integers, floats and doubles and so on, And the difference between typing 10 and typing 10.0 in your code. But on the other hand kosscript hides type information from the script programmer, trying to mask all that off, making places where it leaks through like this stick out. The kosscript programmer can't say something like "print (Double)x * y" to manually force the type promotion..

There appears to be code to check the magnitude of literals in the code so you can say:
set x to 10000.
and end up with an integer, but if you say
set x to 10000000000000.
you end up with a Double.
Can a similar check be done on all integer operations? If the integer math overflowed, then try it again promoted to doubles?

[marianoapp]

I'll have to look into this (you can also write numbers directly in scientific notation, something like "1.23e35")

[Dunbaratu]

Yeah I can't think of an obvious answer to the problem. I'd be curious to see what you come up with. Making everything a Double is easy, but could cause other potential problems when people expect variables to behave like integers in things like loop indeces.