ffi: move guard in the AbstractCompiler

Signed-off-by: Jean Privat <jean@pryen.org>

diff --git a/src/compiler_ffi.nit b/src/compiler_ffi.nit
index 2ebf76d..b885bc7 100644 (file)
--- a/src/compiler_ffi.nit
+++ b/src/compiler_ffi.nit
@@ -261,6 +261,12 @@ redef class CCompilationUnit
        end
 end
 
+redef class AbstractCompiler
+       # Cache to avoid multiple compilation of NULL values
+       # see FIXME in `MNullableType#compile_extern_helper_functions`
+       private var compiled_null_types = new Array[MNullableType]
+end
+
 redef class AbstractCompilerVisitor
        # Create a `RuntimeVariable` for this C variable originating from C user code
        private fun var_from_c(name: String, mtype: MType): RuntimeVariable
@@ -330,8 +336,8 @@ redef class MNullableType
                # FIXME: This is ugly an broke the separate compilation principle
                # The real function MUST be compiled only once, #define pragma only protect the compiler, not the loader
                # However, I am not sure of the right approach here (eg. week refs are ugly)
-               if is_already_compiled then return
-               is_already_compiled = true
+               if v.compiler.compiled_null_types.has(self) then return
+               v.compiler.compiled_null_types.add self
 
                # Internally, implement internal function
                var nitni_visitor = v.compiler.new_visitor
@@ -344,8 +350,6 @@ redef class MNullableType
                nitni_visitor.add("return ret_for_c;")
                nitni_visitor.add("\}")
        end
-
-       private var is_already_compiled = false # FIXME to remove, show above
 end
 
 redef class MExplicitCall