Type conversion for small BigInts into integers about bigints

[konsumlamm]

This issue should be reopened, toSignedInt only provides conversion to signed integers. I'd also argue that it should be renamed to toInt and also support unsigned integers (rather than adding a new toUnsignedInt function).

[konsumlamm]

The only workaround yet is to use the keyword cast[int]. But this does not always do what we want.

That never does what you want. I suggest you read the documentation for cast, it merely reinterprets the underlying bits.

I generally agree that functions to convert a BigInt to some integer types are useful, but how should we handle the case of the BigInt being out of range? I see several possibilities:

• raise an exception
• return an Option
• return the lowest/highest value (especially for floats, where we have Inf/-Inf)
• cut off the higher bits (i.e. return x mod 2^n where n is the number of bits of the resut)

EDIT: Added a 4th option.

[narimiran]

@Araq: you might be interested in this discussion, i.e. give your opinion on how to best proceed with this.

[dlesnoff]

We can not return the lowest/highest value for type conversions to integers as a BigInt can be any value that an int can contain.
The first solution forces the user to wrap any potentially dangerous cast in a try .. except .. finally clause.
I have not used Option in my programs yet, if I understand the user will have to check (everytime ?) whether the Option contains a value ?

[Araq]

IMO the prototype should be:

proc toFixedInt*[T](b: BigInt): Option[T]

with the hopefully obvious semantics.

[Araq]

I also really need pred and succ. Implementation can be as simple as:

proc pred*(a: BigInt; def = 1): BigInt =
  a - initBigInt(def)

proc succ*(a: BigInt; def = 1): BigInt =
  a + initBigInt(def)