[julia-users] scope of variable in parallel reduction ... change from 0.4 to master: bug or feature?

[Andrea Pagnani]

I find the following difference from 0.4 and master. Consider the simple 
function: 

```
function test()
    res = @parallel hcat for i = 1:10
        if true
            i
        end
    end
    res

end
```

on julia0.4 I get

julia> test()
1x10 Array{Int64,2}:
 1  2  3  4  5  6  7  8  9  10

which I think should be the desired output, while on master on master

julia> test()
1x10 Array{Void,2}:
 nothing  nothing  nothing  nothing  nothing  nothing  nothing  nothing 
 nothing  nothing

What do you think?