Documentation: What is an Int?

[expipiplus1]

I've been unable to determine what is representable in a value of type Int in Elm.

This page seems to suggest that Int is a 32bit two's complement integer http://learnyouanelm.github.io/pages/03-types.html

Playing around in the REPL suggests that Int is actually represented as a double precision float. 2^53 is the largest value such that every positive integer below it is representable.

[sgraf812]

Also, the usual weird JavaScript hacks exist if you are concerned about precision. I'm making use of this in intdict, requiring the user to make sure they are actually passing valid integers (yuck!).

Perhaps something like this is worthwhile to have in core?

[sgraf812]

Also, the usual weird JavaScript hacks exist if you are concerned about precision. I'm making use of this in intdict, requiring the user to make sure they are actually passing valid integers (yuck!).

Perhaps something like this is worthwhile to have in core?

[evancz]

Because Elm targets JavaScript, there are certain problems with integers that we cannot fix without very severe performance costs. An Elm Int should be a 32-bit integer, but there is no such thing in JavaScript. So you can think of this as the compiler being non-compliant. The problem is that compliance is basically impossible until we are targeting WebAssembly directly, or some other platform.

One of my goals with Elm is that we should never willingly inherit silly JavaScript problems. This is a case where the language is designed aspirationally, so when the day comes that we can have real integers (through whatever means) Elm does not end up being dumb.

Does that answer your question? Also, I don't know where this should be documented. If someone can find a nice place, maybe it makes sense to do.

[evancz]

Because Elm targets JavaScript, there are certain problems with integers that we cannot fix without very severe performance costs. An Elm Int should be a 32-bit integer, but there is no such thing in JavaScript. So you can think of this as the compiler being non-compliant. The problem is that compliance is basically impossible until we are targeting WebAssembly directly, or some other platform.

One of my goals with Elm is that we should never willingly inherit silly JavaScript problems. This is a case where the language is designed aspirationally, so when the day comes that we can have real integers (through whatever means) Elm does not end up being dumb.

Does that answer your question? Also, I don't know where this should be documented. If someone can find a nice place, maybe it makes sense to do.

[expipiplus1]

Would it be possible to keep this issue open until this information finds a place in the documentation?

[expipiplus1]

Would it be possible to keep this issue open until this information finds a place in the documentation?

[lukewestby]

This sounds like a good addition to http://faq.elm-community.org

[lukewestby]

This sounds like a good addition to http://faq.elm-community.org