Can’t be right, right?
julia> 1.0:5 .- 0.5
1.0:1.0:4.0
julia> collect(1.0:5) .- 0.5
5-element Array{Float64,1}:
0.5
1.5
2.5
3.5
4.5
Try adding parentheses around the range in the first example — it’s a precedence thing.
Yes, that’s it but quite unexpected, at least to me.
2 Likes
If you’re coming from MATLAB, e.g., this is the very expected thing!
Agreed. I’ve been caught on it too. That’s how I knew 
1 Like
I must confess. I have many years of Matlab and, damn it, I had never noticed that. Who knows why, probably there I never missed the parentheses but here in Julia I was expecting a different thing.
Yeah, I generally end up adding parentheses for complicated ranges because the precedence isn’t always obvious.
On the other hand, the current behavior allows you to do:
for i in 1:N-1
...
end
which feels pretty natural.
3 Likes
Presumably you’ve written 1:n-1 or a+1:n?
1 Like
I like the Fortress idea where whitespace which is inconsistent with precedence is a syntax error: this would have caught the original example here.
See also Non-transitive operator precedence · Issue #18714 · JuliaLang/julia · GitHub
10 Likes
Yes, I really like that too. It would be breaking though so it would have to be opt-in or wait until 2.0.
No. I can’t reconstruct it anymore but it involved something like this
julia> x=1:5;
julia> x - 0.5
┌ Warning: `a::AbstractArray - b::Number` is deprecated, use `a .- b` instead.
│ caller = top-level scope at none:0
└ @ Core none:0
0.5:1.0:4.5
which fixed in my head the array array side of this and then later using 1:5 .- 0.5 and being surprised that the results were different. Being late at night might had influenced too.