replicate gamma(0.0) and gamma(-0.0) behaviour

Here it's possible to tell which of +Inf and -Inf to use

Author: MikaelSlevinsky

src/confluent.jl

@@ -6,11 +6,11 @@ Compute Kummer's confluent hypergeometric function `M(a, b, z) = ₁F₁(a; b; z
 function _₁F₁(a, b, z; kwds...)
     if isequal(a, b) # 13.6.1
         return exp(z)
-    elseif -b ∈ ℕ₀
+    elseif -b ∈ ℕ
         if -a ∈ ℕ₀ && real(a) ≥ real(b)
             return _₁F₁maclaurin(a, b, z; kwds...)
         else
-            return throw(DomainError(b, "M(a, b, z) = ₁F₁(a; b; z) is not defined for nonpositive integer b unless a is also a nonpositive integer and a ≥ b."))
+            return throw(DomainError(b, "M(a, b, z) = ₁F₁(a; b; z) is not defined for negative integer b unless a is a nonpositive integer and a ≥ b."))
         end
     elseif -a ∈ ℕ₀
         return _₁F₁maclaurin(a, b, z; kwds...)

test/runtests.jl

@@ -466,6 +466,8 @@ end
 @testset "M" begin
     @test M(-3, -3, 0.5) ≡ exp(0.5)
     @test M(0, -1, 10) ≡ 1.0
+    @test M(1.2,  0.0, 0.1) ==  Inf # Mimick gamma( 0.0) =  Inf
+    @test M(1.2, -0.0, 0.1) == -Inf # and    gamma(-0.0) = -Inf
     @test_throws DomainError M(1, -2, 0.5)
     @test_throws DomainError M(-3, -2, 0.5)
     @test M(-2, -3, 0.5) ≡ 1.375