abstract_compiler: Replace the use of `n_float` by the `value`

Signed-off-by: Florian Deljarry <deljarry.florian@gmail.com>

diff --git a/src/compiler/abstract_compiler.nit b/src/compiler/abstract_compiler.nit
index c74933c..3fc2f40 100644 (file)
--- a/src/compiler/abstract_compiler.nit
+++ b/src/compiler/abstract_compiler.nit
@@ -1776,10 +1776,10 @@ abstract class AbstractCompilerVisitor
        # Generate a float value
        #
        # FIXME pass a Float, not a string
-       fun float_instance(value: String): RuntimeVariable
+       fun float_instance(value: Float): RuntimeVariable
        do
                var t = mmodule.float_type
-               var res = new RuntimeVariable("{value}", t, t)
+               var res = new RuntimeVariable("{value.to_hexa_exponential_notation}", t, t)
                return res
        end
 
@@ -3820,7 +3820,7 @@ redef class AIntegerExpr
 end
 
 redef class AFloatExpr
-       redef fun expr(v) do return v.float_instance("{self.n_float.text}") # FIXME use value, not n_float
+       redef fun expr(v) do return v.float_instance(self.value.as(Float))
 end
 
 redef class ACharExpr