Static

Static defines a set of types (True, False, StaticInt{value::Int}, StaticFloat64{value::Float64}, StaticSymbol{value::Symbol}) that may be dispatched on (similar to Base.Val{value}). Unlike Base.Val, instances of these types provide "static" values (meaning known at compile time) that in many cases work interchangeably with dynamic values (meaning the value is known at runtime but not compile time). This is particularly useful when designing types whose fields may be dynamically or statically known, such as a range whose size may be dynamic or statically known.

Generic conversions to static values, dynamic values, and compile-time-known values are accomplished with the methods static, dynamic, and known, respectively.

julia> using Static

julia> static(1)
static(1)

julia> dynamic(static(1))
1

julia> dynamic(1)
1

julia> typeof(static(1))
StaticInt{1}

julia> known(typeof(static(1)))
1

julia> known(typeof(1)) === nothing  # `Int` has no compile-time-known value
true

Types, Dispatch, Compile-Time, and StaticNumbers.jl

Static.jl does not subtype the Base number types. For example, !(StaticInt <: Integer). The reason for this is that it invalidates all downstream compilation caches. This has a major effect, for example, causing LoopVectorization.jl to not precompile and ultimately giving the ODE solvers >20 second extra compile times. To avoid these invalidations, the dispatches were removed. An alternative library StaticNumbers.jl is more ergonomic but has these invalidations. Use the library that is more appropriate for your use case.