bugfix for large args

Author: JeffreySarnoff

src/math/elementary/hyptrig.jl

@@ -1,5 +1,5 @@
 function sinh(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return x
+    HI(x) >= 709.0 && return DoubleFloat{T}(copysign(Inf, HI(x)))
     x < 0 && return -sinh(-x)
     iszero(x) && return zero(x)
     !isfinite(x) && return nan(typeof(x))
@@ -10,7 +10,7 @@ function sinh(x::DoubleFloat{T}) where {T<:IEEEFloat}
 end
 
 function cosh(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return abs(x)
+    HI(abs(x)) >= 709.0 && return DoubleFloat{T}(Inf)
     x < 0 && return cosh(-x)
     iszero(x) && return one(x)
     !isfinite(x) && return nan(typeof(x))
@@ -21,7 +21,7 @@ function cosh(x::DoubleFloat{T}) where {T<:IEEEFloat}
 end
 
 function tanh(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return copysign(one(DoubleFloat{T}), x)
+    HI(abs(x)) >= 709.0 && return copysign(one(DoubleFloat{T}), x)
     return sinh(x) / cosh(x)
 end
 #=
@@ -37,16 +37,16 @@ end
 =#
 
 function csch(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return copysign(zero(DoubleFloat{T}), x)
+    HI(abs(x)) >= 709.0 && return copysign(zero(DoubleFloat{T}), x)
     return inv(sinh(x))
 end
 
 function sech(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return zero(DoubleFloat{T})    
+    HI(abs(x)) >= 709.0 && return zero(DoubleFloat{T})    
     return inv(cosh(x))
 end
 
 function coth(x::DoubleFloat{T}) where {T<:IEEEFloat}
-    isinf(x) && return copysign(one(DoubleFloat{T}), x)
+     HI(abs(x)) >= 709.0 && return copysign(one(DoubleFloat{T}), x)
     return cosh(x) / sinh(x)
 end