On Tuesday, April 15, 2014 1:14:16 PM UTC-4, Toivo Henningsson wrote:
>
> Actually, it seems like it already works :) I didn't know that.
> Also, the
> <http://docs.julialang.org/en/release-0.2/manual/mathematical-operations/#chaining-comparisons>
> manual<http://docs.julialang.org/en/release-0.2/manual/mathematical-operations/#chaining-comparisons>says
> that && is used to chain dotless comparisons and & to chain dotted
> ones. I only wonder what is used if you try a combination such as 1 .< 2 >
> 0 (which evaluates to true).
>
It doesn't work by translating into .<(a,b,c) for exactly that reason.
It's pretty cool:
julia> dump(:(1 .< 2 > 0))
Expr
head: Symbol comparison
args: Array(Any,(5,))
1: Int64 1
2: Symbol .<
3: Int64 2
4: Symbol >
5: Int64 0
typ: Any
julia> f() = 1 .< 2 > 0
f (generic function with 1 method)
julia> code_lowered(f,())
1-element Array{Any,1}:
:($(Expr(:lambda, {}, {{},{},{}}, :(begin # none, line 1:
return (1 .< 2) & (2 > 0)
end))))
It's very impressive, especially if you add another dotless comparator
(since then it effectively uses &&, with control flow implications).
julia> g() = 1 .< 2 > 0 < 1
g (generic function with 1 method)
julia> code_lowered(g,())
1-element Array{Any,1}:
:($(Expr(:lambda, {}, {{:#s31},{{:#s31,:Any,2}},{}}, :(begin # none, line
1:
unless 2 > 0 goto 0
#s31 = 0 < 1
goto 1
0:
#s31 = false
1:
return (1 .< 2) & #s31
end))))
