I’ve decided that I agree with you, and I’ve updated the spec and examples for block expressions accordingly. In block mode the commas are now optional, similar to how Vectors work.
I’m also playing around more with functor composition, and it’s looking super clean.
I've updated the demo code to reflect these changes here, as well as supporting some more (though still not all) broadcasting cases.
see github https://github.com/uniment/ChainingDemo.jl
Here are updated demo examples.
See github https://github.com/uniment/ChainingDemo.jl
Also, here are some new demos that work now:
demo" (1,2,3)--abs2.(_) "
demo" (_+2+3) "
demo" (_-2-3) "
demo" 1--(_+2+3) "
demo" 1--(_-2-3) "
demo" tan(sin(cos(_)))--acos(asin(atan(_))) "
demo" tan(sin(cos(_)))--acos(asin(atan(_))) "(1)
demo" cos(_)--sin--tan--atan--asin--acos "
demo" cos(_)--sin--tan--atan--asin--acos "(1)
demo" --(cos; sin; tan; atan; asin; acos) "
demo" --(cos; sin; tan; atan; asin; acos) "(1)
demo" --cos--sin--tan--atan--asin--acos "
demo" --cos--sin--tan--atan--asin--acos "(1)
# cos--sin--tan--atan--asin--acos doesn't work,
# because sin(::Function) is not defined.
# Should it be though? 🤔 When a method for f(x::T) isn't found,
# if T is a callable type, should f∘x be returned?
demo" (a=1,)--(_.a+1; it+it^2) "
demo" ((1,2),(2,3),(3,4)).--(_[1]^2 + 1) "
demo" 1--((1,2,3)[_+1]) "
demo" (0:10)--filter(isodd, _)--map(_/2+1, _) "
demo" [(a=i,) for i=0:10]--filter(_.a%3==0, _) "
demo"""
"a=1 b=2 c=3"--begin
split
it.--(split(_,"="); (it[1]--Symbol => parse(Int,it[2])))
NamedTuple
end
"""
using DataFrames
df = DataFrame(id=1:100, group=rand(1:5, 100), age=rand(1:30, 100));
demo"""
df--begin
dropmissing
filter(:id => (_%5==0), _)
groupby(_, :group)
combine(_, :age => sum)
end
"""
demo" (0:0.25:1).--atan(cos(π*_)) "
demo" [1,2,3]--((l=length(it); it); sum(it)/l) "
demo" 1--(it+1, it-2) "
demo" 1--[it+1, it-2] "
demo" 1--Dict(:a=>it+1, :b=>it-2) "
demo" 1--Set((it+1, it-2)) "
demo" (1,2,3)--[i^2 for i ∈ it] "
demo" (1,2,3)--[(i*_)^2 for i ∈ it] "