Python has it, plus a general facility for writing your own "context manager"s. Common Lisp has had it forever. I would be surprised if there are not several other languages as well. There’s a tendency to attribute innovation to the big player (Not just in software, but engineering in general, and probably wider still) I’m susceptible to this. I kind of vaguely thought Julia got comprehensions from python.
1 Like
Using unicode symbols for constants. Instead of calling something like np.pi writing π just feels right.
5 Likes
I like end
As in xs[end] for the last element.
but end can do lots of other cool things.
Because it is replaces with lastindex(xs) during lowering
Meaning you can use it just like a number.
Like getting the first 5 elements, or all the elements if there are less than 5:
xs[1:min(end, 5)]
Or or breaking the data into splits:
train = data[1:9end÷10]
test = data[9end÷10 + 1 : end]
27 Likes
I like @mbauman 's example for constructing from a finite array an array whose elements repeat forever x[mod1(i, end)].
5 Likes
As of now my favorite sugar is
julia> function f()
(; a = 1, b = 2, c = 3)
end
f (generic function with 1 method)
julia> (; b, c) = f()
(a = 1, b = 2, c = 3)
julia> b
2
julia> c
3
16 Likes
This had never occured to me - really neat!
This is erroring for me in the REPL on 1.6rc1. Any idea why?
ERROR: syntax: invalid assignment location "; b, c" around REPL[9]:1
Stacktrace:
[1] top-level scope
@ REPL[9]:1
1 Like
It’s 1.7 feature 
For those, who like to live on the edge.
10 Likes
Well past the edge, since 1.6 isn’t out yet perhaps we could call it “off the edge of the world and into the starry darkness of nightly”
5 Likes
Also wrong on 1.5.3. I am wondering what the feature is called and where is the documentation 
It’s worth noting that this new feature also allows you to unpack structs (i.e., pattern matching) in function arguments, like this:
struct A
x
y
end
foo((; x, y)::A) = x + y
julia> foo(A(1, 2))
3
22 Likes
This (; ...) named destructuring is awesome for working with data frames:
julia> using DataFrames
julia> df = DataFrame(rand.([Bool, "abc", Float64, Float64], 5),
["flag", "type", "x", "y"])
5×4 DataFrame
Row │ flag type x y
│ Bool Char Float64 Float64
─────┼──────────────────────────────────
1 │ false c 0.757074 0.637436
2 │ true a 0.12791 0.163183
3 │ false b 0.452078 0.547038
4 │ true c 0.187923 0.0529017
5 │ true b 0.10274 0.242524
julia> for (; x, y) in eachrow(df)
println("x-y = $(x-y)")
end
julia> map(eachrow(df)) do (; x, y)
x-y
end
julia> (; x, y) = df
17 Likes
No body mention function definition?
e.g. f(x) = x^2
19 Likes
I wonder if anyone here realized that refined sugar is poisonous to humans?
9 Likes
Some poison from excess sugar:
julia> f(x) = f(x)
f (generic function with 1 method)
julia> f(1)
ERROR: StackOverflowError:
Stacktrace:
[1] f(x::Int64) (repeats 79984 times)
7 Likes
Thank you, I would rather take the other 364 days without sugar.
2 Likes
Every feature of Julia that I don’t yet know is a potential favorite syntactical sugar.
Just a moments ago, my favorite changed from the |> and .|> to the assert macro
@assert( σ >=0 , "σ must be non-negative")
2 Likes