Possible bug: StackOverflowError with inds::Tuple{Int32,Colon}

[toollu]

Vec64  = zeros(6)
Vec32  = Float32.(zeros(6))
ind_a_64 = 3
ind_a_32 =Int32.(ind_a_64)

julia> reshape(Vec64,ind_a_64,:)
3×2 Matrix{Float64}:
 0.0  0.0
 0.0  0.0
 0.0  0.0

julia> reshape(Vec32,ind_a_64,:)
3×2 Matrix{Float32}:
 0.0  0.0
 0.0  0.0
 0.0  0.0

julia> reshape(Vec64,ind_a_32,:)
ERROR: StackOverflowError:
Stacktrace:
 [1] reshape(A::Vector{Float64}, inds::Tuple{Int32, Colon}) (repeats 79984 times)
   @ OffsetArrays ~/.julia/packages/OffsetArrays/5zoKM/src/OffsetArrays.jl:306

julia> reshape(Vec32,ind_a_32,:)
ERROR: StackOverflowError:
Stacktrace:
 [1] reshape(A::Vector{Float32}, inds::Tuple{Int32, Colon}) (repeats 79984 times)
   @ OffsetArrays ~/.julia/packages/OffsetArrays/5zoKM/src/OffsetArrays.jl:306

julia> reshape(Vec32,ind_a_32,2)
3×2 Matrix{Float32}:
 0.0  0.0
 0.0  0.0
 0.0  0.0

julia> VERSION
v"1.6.0"

[johnnychen94]

Julia explicitly uses Int as index type (See Base.Dims). For some indices-related functions, there might be fallback methods for Integer types, but many functions don't.

julia> x = rand(8, 8);

julia> reshape(x, 4, :)
4×16 Matrix{Float64}:
 0.690637  0.555645  0.821886  0.406586  0.539375   …  0.0998133  0.546501  0.913381  0.778848
 0.189625  0.559087  0.849438  0.163502  0.0132832     0.769946   0.951078  0.20459   0.38723
 0.901638  0.48459   0.471441  0.440043  0.256732      0.23922    0.506384  0.482037  0.120117
 0.377634  0.161835  0.841939  0.722828  0.877613      0.0740276  0.066585  0.820278  0.807347

julia> reshape(x, Int32(4), :)
ERROR: MethodError: no method matching reshape(::Matrix{Float64}, ::Int32, ::Colon)

The StackOverFlow is not good, so definitely needs improvement.