Array type init with a negative value.

[kalwalt]

If you create an Array like this one: var test: Array<byte> = Array<byte>.new(-1); with a negative length it will compile wthout errors but you will receive a !LengthCheckException when you run the builded app. Does it is correct? Because the parameter in the new constructor is an int but should be better an unsigned int type instead?

[titzer]

The exception is the correct behavior. I agree that it would be better for Array to take an unsigned integer, but in the very beginning there weren't any unsigned types. In the future I was hoping that this could be improved, and that arrays could also accept larger-sized integer types.

[kalwalt]

The exception is the correct behavior. I agree that it would be better for Array to take an unsigned integer, but in the very beginning there weren't any unsigned types. In the future I was hoping that this could be improved, and that arrays could also accept larger-sized integer types.

Sorry to re-open this issue, i want only to answer. I agree with you that this is the correct behavior, I was only thinking if it can be improved. Not a real issue, to be honest i will never assign a negative value to an Array.new(), but if you have a function that as a parameter an unsigned integer and this is assigned to the Array you need to cast to int. I hope that in a future this can be improved.

[titzer]

Sure. Not that array accesses do allow using an integer of any width (or signedness), and if it's a signed integer, the bounds check will check for negative values as well. So this code works:

var x = Array<int>.new(33);
var y1 = x[0uL]; // u64 index
var y2 = x[0L]; // i64 index
var y3 = x[0u]; // u32 index

In theory, new could also be polymorphic in this way.

A natural question is then: what is the type of Array.length--right now it's actually int, but guaranteed to be non-negative. So it's kind of u31. But Virgil's other rules for arithmetic would make it inconvenient to make the type of an array length u31.

Sorry for the small ugliness here.

[kalwalt]

Thank you for the clarification Ben @titzer !