use throw(ArgumentError) instead of error in distribution constructors

src/univariate/continuous/arcsine.jl

@@ -3,7 +3,7 @@ immutable Arcsine <: ContinuousUnivariateDistribution
     b::Float64
 
     function Arcsine(a::Float64, b::Float64)
-        a < b || throw(ArgumentError("a must be less than b."))
+        a < b || throw(ArgumentError("Arcsine: a must be less than b."))
         new(a, b)
     end
 

src/univariate/continuous/beta.jl

@@ -4,7 +4,7 @@ immutable Beta <: ContinuousUnivariateDistribution
 
     function Beta(α::Real, β::Real)
         (α > zero(α) && β > zero(β)) ||
-            throw(ArgumentError("α and β must be positive"))
+            throw(ArgumentError("Beta: α and β must be positive"))
         @compat new(Float64(α), Float64(β))
     end
 

src/univariate/continuous/betaprime.jl

@@ -3,7 +3,8 @@ immutable BetaPrime <: ContinuousUnivariateDistribution
     β::Float64
 
     function BetaPrime(α::Float64, β::Float64)
-        (α > zero(α) && β > zero(β)) || error("α and β must be positive")
+        (α > zero(α) && β > zero(β)) ||
+            throw(ArgumentError("BetaPrime: α and β must be positive."))
         @compat new(Float64(α), Float64(β))
     end
 

src/univariate/continuous/biweight.jl

@@ -3,7 +3,8 @@ immutable Biweight <: ContinuousUnivariateDistribution
     σ::Float64
 
     function Biweight(μ::Real, σ::Real)
-        σ > zero(σ) || throw(ArgumentError("scale must be positive"))
+        σ > zero(σ) ||
+            throw(ArgumentError("Biweight: σ must be positive."))
         @compat new(Float64(μ), Float64(σ))
     end
 

src/univariate/continuous/cauchy.jl

@@ -3,7 +3,8 @@ immutable Cauchy <: ContinuousUnivariateDistribution
     σ::Float64
 
     function Cauchy(μ::Real, σ::Real)
-        σ > zero(σ) || error("Cauchy: scale must be positive")
+        σ > zero(σ) ||
+            throw(ArgumentError("Cauchy: σ must be positive."))
         @compat new(Float64(μ), Float64(σ))
     end