Saving a array of composite type Julia

Question

For an experiment I am running it is handy for me to define my own types. Tha problem is that the computation is really heavy so I need to run it on a server and then access the data to plot results.

So the first type has certain fields that I define and then the other types do not have a fixed number of fields. Basically I have this kind of intertwined types:

type TestType1
    value::Int
end

type TestType2
    testvec::Any
    TestType2()=new(Test2[])
end

type TestType3
    testtype2::Any
    TestType3()=new(TestType2[])
end

type TestType4
    testtype3::Any
    TestType4()=new(TestType3[])
end

Because the TestType1 never changes but within TestType2 I want to store different number of testtype1 values etc...

So the chain creation work like that :

test1=Test2(2);
test2=TestType2();
test3=TestType3();
test4=TestType4();
push!(test2.testvec,test1)
push!(test3.testtype2,test2)
push!(test4.testtype3,test3)

This works well I can have an easy access to everything I need in my experiment.

although the problem is as I run it on a server I need to be able to store test4 at the end of the run and then open it again on an other server to use the datas and do nice plots.

I tryed to use JLD but I think it is for dictionaries and it does not allow me to retrieve the information.

I would like to be able to call all the fields inside my test4 file stored somewhere the same way I do without changing server.

I am quite new in Julia and in OOP in general I dont have any idea idea how to do what I want to do.

Can anyone help please?

Thank you!

Answer 1

JLD.load returns a dictionary, inside it is your object. For example:

# test_types.jl

__precompile__()

module TestTypes

using JLD: save

import Base: push!

export TestType1, TestType2, TestType3, TestType4

struct TestType1
    value::Int
end

struct TestType2
    data::Vector{TestType1}

    TestType2() = new(TestType1[])
end

struct TestType3
    data::Vector{TestType2}

    TestType3() = new(TestType2[])
end

struct TestType4
    data::Vector{TestType3}

    TestType4() = new(TestType3[])
end

for n in 2:4
    @eval function Base.push!(test_type::$(Symbol(:TestType, n)), test_arg::$(Symbol(:TestType, n-1)))
        push!(test_type.data, test_arg)
    end
end

function main()
    test1 = TestType1(2)
    test2 = TestType2()
    test3 = TestType3()
    test4 = TestType4()

    push!(test2, test1)
    push!(test3, test2)
    push!(test4, test3)

    file = "test_resutls.jld"
    save(file, "test4", test4)
    info("Test results saved in: $file")
    return @show test4
end

end

An interaction looks like this:

julia> include("test_types.jl")
Main.TestTypes

julia> using Main.TestTypes: main

julia> original_test4 = main();
INFO: Test results saved in: test_resutls.jld
test4 = Main.TestTypes.TestType4(Main.TestTypes.TestType3[Main.TestTypes.TestType3(Main.TestTypes.TestType2[Main.TestTypes.TestType2(
Main.TestTypes.TestType1[Main.TestTypes.TestType1(2)])])])

julia> using JLD: load

julia> results = load("test_resutls.jld")
Dict{String,Any} with 1 entry:
  "test4" => Main.TestTypes.TestType4(Main.TestTypes.TestType3[Main.TestTypes.TestType3(Main.TestTypes.TestType2[Main.TestTypes.Test…


julia> copy_test4 = results["test4"]  # your data is inside this dict!
Main.TestTypes.TestType4(Main.TestTypes.TestType3[Main.TestTypes.TestType3(Main.TestTypes.TestType2[Main.TestTypes.TestType2(Main.Tes
tTypes.TestType1[Main.TestTypes.TestType1(2)])])])

julia> original_test4.data[1].data[1].data[1].value == copy_test4.data[1].data[1].data[1].value
true