Which F# type names and declaration syntax are considered idiomatic?

Question

F# has multiple ways to declare the same types. This is likely because of the dual lineage of F# as both a member of the ML family and a .NET language. I haven't been able to find any guidance on which style is more idiomatic.

Specifically, I want to know:

• Which is more idiomatic for 64-bit IEEE 754 floating-point numbers in F#, float or double?
• Which is a more idiomatic way in F# to declare an array type:
  • int[]
  • int array
  • array<int>

Sources:

• https://learn.microsoft.com/dotnet/fsharp/language-reference/basic-types
• https://learn.microsoft.com/dotnet/fsharp/language-reference/fsharp-types#syntax-for-types

Context: I'm working on some API documentation that is explaining how data in a data store maps to .NET types, along with how those types are typically declared in both C# and F#.

Answer 1

For doubles, it's pretty much always float. Unless you deal with both singles and doubles and need to ensure clarity I guess.

For generic types, the usual syntax I use and see people use is:

• int option
• int list
• int[]
• For all other types, including F#-specific ones like Async, Set, and Map, angle bracket syntax is used.

The only type that I feel has a significant split is seq (an alias for IEnumerable): I'd say the majority of people use seq<int> but a significant number of people write int seq. Either way, you should definitely use seq and not IEnumerable. Similarly, you should use the alias ResizeArray for System.Collections.Generic.List.

Answer 2

The F# Core Library reference, which seems like a good example to follow, seems to prefer float, int[] and seq<int>.