SVector{N,T} only accept Int64 but not Int32

[Blumenkranz]

Maybe Tuple{Vararg{Union{StaticArraysCore.Dynamic, Int64}}} can be changed to Tuple{Vararg{Union{StaticArraysCore.Dynamic, Integer}}}? When I use C API, all integers are Cint, i.e. Int32, which requires extra type conversion.

[mattsignorelli]

I also ran into this problem, unfortunately wasting two hours of my work to find this was the cause.

All throughout the StaticArrays package, ::Int is used where ::Integer should be allowed, just like all AbstractArrays in Julia. Or at the least, a descriptive error should be printed stating Int must be specifically used.

julia> length(MVector{6})
6

julia> length(MVector{Cint(6)})
ERROR: TypeError: in typeassert, expected Tuple{Vararg{Union{StaticArraysCore.Dynamic, Int64}}}, got a value of type Tuple{Int32}
Stacktrace:
 [1] Size{(6,)}()
   @ StaticArraysCore ~/.julia/packages/StaticArraysCore/7xxEJ/src/StaticArraysCore.jl:488
 [2] Size(::Type{Tuple{6}})
   @ StaticArraysCore ~/.julia/packages/StaticArraysCore/7xxEJ/src/StaticArraysCore.jl:495
 [3] Size(::Type{MVector{6}})
   @ StaticArraysCore ~/.julia/packages/StaticArraysCore/7xxEJ/src/StaticArraysCore.jl:517
 [4] length(a::Type{MVector{6}})
   @ StaticArrays ~/.julia/packages/StaticArrays/xEhFV/src/abstractarray.jl:2
 [5] top-level scope
   @ REPL[9]:1

[nsajko]

Relaxing Int to Integer is not really an option. Consider this:

julia> Val{Int32(7)} == Val{Int64(7)}
false

Or at the least, a descriptive error should be printed stating Int must be specifically used.

That's the current error message IMO. Do you have a specific idea on how to improve it?

[nsajko]

When I use C API, all integers are Cint, i.e. Int32, which requires extra type conversion.

In what way is that an issue? Does the conversion not constant fold?

[nsajko]

Opened an upstream issue: https://github.com/JuliaLang/julia/issues/57189

[nsajko]

duplicate of #798

[BSchilperoort]

Encountered this issue as well, and the error message in no way hinted that the problem was in the Int32 type...

LoadError: ArgumentError: cannot reinterpret `Float32` as `StaticArraysCore.SVector{4, Float32}`, 
    type `StaticArraysCore.SVector{4, Float32}` is not a bits type

Where 4 is a Int32 and not Int64. But that is not explicitly shown due to https://github.com/JuliaLang/julia/issues/57189

unfortunately wasting two hours of my work to find this was the cause.

+ another two hours wasted here 🙄

---

To reproduce;

using StaticArrays

x = ones(8)
sv = SVector{4, Float64}
reinterpret(sv, x)  # runs fine

sv = SVector{Int32(4), Float64}  # no error here
reinterpret(sv, x)  # "not a bits type" error