Shared array for non bit types

New to Julia
1

Probably this is a silly question but I couldn’t find any solution for that.
I want to iterate over a vector of my own type and do some calculation.
Here is a MWE

# julia -p 5
@everywhere mutable struct MyType
    val::Int64
end

function check_move(v::Vector{MyType})
    @sync @distributed for i in 1:length(v)
    # for i in 1:length(v)
        sleep(1)
        v[i].val = 100
    end
end

# initialize the vector of my own type
my_vec = Vector{MyType}(undef,5)
for i in 1:5
    my_vec[i] = MyType(i)
end

# try to change each one of the values to 100
@time check_move(my_vec)
julia> my_vec
# 5-element Vector{MyType}:
#  MyType(1)
#  MyType(2)
#  MyType(3)
#  MyType(4)
#  MyType(5)

But it should be a vector with MyType(100), why does this happen?
Thanks,
Elyco

2

You need to use DistributedArrays for this. At the moment you have not done anything here to facilitate interprocess communication.

Can you use threads instead? Threads can address the same memory space (e.g. an array) and thus do not need as much communication.