RFC: lower x^literal as x^Val{literal} for integer literals (#20530)

lower x^literal as x^Val{literal} for integer literals (closes #20527)

NEWS.md

@@ -73,6 +73,10 @@ Language changes
  * The `typealias` keyword is deprecated, and should be replaced with
  `Vector{T} = Array{T,1}` or a `const` assignment.
 
+ * Experimental feature: `x^n` for integer literals `n` (e.g. `x^3`
+ or `x^-3`) is now lowered to `x^Val{n}`, to enable compile-time
+ specialization for literal integer exponents ([#20530]).
+
 Breaking changes
 ----------------
 

base/intfuncs.jl

@@ -195,6 +195,11 @@ end
 ^(x::Number, p::Integer) = power_by_squaring(x,p)
 ^(x, p::Integer) = power_by_squaring(x,p)
 
+# x^p for any literal integer p is lowered to x^Val{p},
+# to enable compile-time optimizations specialized to p.
+# However, we still need a fallback that calls the general ^:
+^{p}(x, ::Type{Val{p}}) = x^p
+
 # b^p mod m
 
 """

base/promotion.jl

@@ -253,6 +253,11 @@ end
 
 Exponentiation operator. If `x` is a matrix, computes matrix exponentiation.
 
+If `y` is an `Int` literal (e.g. `2` in `x^2` or `-3` in `x^-3`), the Julia code
+`x^y` is transformed by the compiler to `x^Val{y}`, to enable compile-time
+specialization on the value of the exponent. (As a default fallback,
+however, `x^Val{y}` simply calls the `^(x,y)` function.)
+
 ```jldoctest
 julia> 3^5
 243

src/julia-syntax.scm

@@ -2057,6 +2057,10 @@
  (expand-forms
  `(call (core _apply) ,f ,@(tuple-wrap argl '())))))
 
+ ((and (eq? f '^) (length= e 4) (integer? (cadddr e)))
+ (expand-forms
+ `(call ^ ,(caddr e) (call (core apply_type) (top Val) ,(cadddr e)))))
+
  ((and (eq? f '*) (length= e 4))
  (expand-transposed-op
  e

test/numbers.jl

@@ -2902,6 +2902,18 @@ end
  @test rem2pi(T(-4), RoundUp) == -4
 end
 
+import Base.^
+immutable PR20530; end
+^(::PR20530, p::Int) = 1
+^{p}(::PR20530, ::Type{Val{p}}) = 2
+@testset "literal powers" begin
+ x = PR20530()
+ p = 2
+ @test x^p == 1
+ @test x^2 == 2
+ @test [x,x,x].^2 == [2,2,2]
+end
+
 @testset "iszero" begin
  # Numeric scalars
  for T in [Float16, Float32, Float64, BigFloat,