Tutorial 6: Extra Tips and Tricks

Some quick tips that we’ve learnt the hard way…

1. In the REPL, you can use the up arrow to scroll through past code
2. You can even filter through your past code by typing the first letter of a line previously executed in the REPL. For example, try typing p in the REPL and using the up arrow to scroll through your code history, you should quickly find the last plot command you executed.
3. Toggle word wrap via View>Toggle Word Wrap or alt-Z
4. To interrupt Julia while it is running through code use ⌘/Ctrl + C
5. Red wavy line under code in your script = error in code
6. Blue wavy line under code in your script = possible error in code
7. Errors and possible errors can be viewed in the PROBLEMS section of the REPL
8. You can view your current variables (similar to the top right hand panel in RStudio) by clicking on the Julia explorer: Julia workspace symbol in the activity bar (three circles). You can then look at them in more detail by clicking the sideways arrow (when allowed).
9. The copy function in Julia does not behave as one intuitively expects and if you want to make changes to a copied item without altering the original use deepcopy instead

COPYING versus DEEP COPYING

Julia has a strange copying behaviour where if a = b, any change in b will automatically cause the same change in a. For example, let’s make an array of three numbers:

aa = [1,2,3]

3-element Vector{Int64}:
 1
 2
 3

Here we make a copy of aa and call it bb.

bb = aa
print(bb)

[1, 2, 3]

Now, we replace the second element of bb with the value 41.

bb[2] = 41
bb

3-element Vector{Int64}:
  1
 41
  3

The default behaviour of = in Julia is to UPDATE any copy of the array, so we see the same change in aa:

aa

3-element Vector{Int64}:
  1
 41
  3

Whoa… that’s not what we expect, though it’s pretty cool. This approach is advantageous because it lets Julia save memory, however, it is not ideal.

To remedy this, and create copies that don’t update their parent or other offspring, we can force c to be an independent copy of a using the deepcopy function:

cc = deepcopy(aa)

3-element Vector{Int64}:
  1
 41
  3

cc[3] = 101

101

See how only the third element of cc s changed but not aa?

aa,cc

([1, 41, 3], [1, 41, 101])

9. You can view a .csv or .txt file by clicking on a file name in the project directory (left panel) - this opens a viewing window. CSV’s also have a built in ‘Preview’ mode - try using right click>Open Preview on a .csv file and check it out.

10. Docstrings are a supplement to #-based comments for documenting your workflow. Basically, any string that appears just before an object will be interpreted as documenting it. When you use a docstring it is possible to call that description using help (? in the REPL) or hovering over the object in VSCode. This can be very useful when you start declaring variables or building your own functions.

Docstrings

Below is an examle of a docstring for a global variable as well as how it would look for a function

"a wonderfully documented and described variable"
var = 101

you can (should) also do this with functions

"""
    _rategradient(∂X, ∂Y)

Returns the rate of change in units of the values stored in the grid, and the
angle of the change in wind direction, *i.e.* an angle of 180 means that the
value is increasing *from* the south. When both ∂X and ∂Y are equal to 0, the
angle is assumed to be 0.
"""
function _rategradient(∂X::T, ∂Y::T) where {T <: Number}
    if ∂X == ∂Y == 0.0
        return (0.0, 0.0)
    end
    m = sqrt(∂X^2 + ∂Y^2)
    Δ = ∂X >= 0.0 ? 0.0 : 180.0
    θ = rad2deg(atan(∂X, ∂Y)) + Δ
    θ = ∂X > 0.0 ? θ + 180.0 : θ
    return (m, θ)
end