Merge pull request #254 from ReactiveBayes/error-hint-determentistic-…

…mapping

Give a hint if a user uses = operator wrong

src/GraphPPL.jl

@@ -50,4 +50,20 @@ macro model(model_specification)
     return esc(GraphPPL.model_macro_interior(DefaultBackend, model_specification))
 end
 
+function __init__()
+    if isdefined(Base.Experimental, :register_error_hint)
+        Base.Experimental.register_error_hint(MethodError) do io, exc, argtypes, kwargs
+            if any(x -> x <: VariableRef, argtypes)
+                print(io, "\nOne of the arguments to ")
+                printstyled(io, "`$(exc.f)`", color = :cyan)
+                print(io, " is of type ")
+                printstyled(io, "`GraphPPL.VariableRef`", color = :cyan)
+                print(io, ". Did you mean to create a new random variable with ")
+                printstyled(io, "`:=`", color = :cyan)
+                print(io, " operator instead?")
+            end
+        end
+    end
+end
+
 end # module

test/model_macro_tests.jl

@@ -2030,3 +2030,16 @@ end
 
     @test !isnothing(GraphPPL.create_model(somemodel(a = 1, b = 2)))
 end
+
+@testitem "model should warn users against incorrect usages of `=` operator with random variables" begin 
+    using GraphPPL, Distributions
+    import GraphPPL: @model
+
+    @model function somemodel()
+        a ~ Normal(0, 1)
+        t = exp(a)
+        y ~ Normal(0, t)
+    end
+
+    @test_throws "One of the arguments to `exp` is of type `GraphPPL.VariableRef`. Did you mean to create a new random variable with `:=` operator instead?" GraphPPL.create_model(somemodel())
+end