bindings: Explicitly pass ValueType to toImplArray

third_party/WebKit/Source/bindings/scripts/v8_types.py

 # Copyright (C) 2013 Google Inc. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions are
 # met:
 #
 #     * Redistributions of source code must retain the above copyright
 # notice, this list of conditions and the following disclaimer.
 #     * Redistributions in binary form must reproduce the above
 # copyright notice, this list of conditions and the following disclaimer
@@ skipping 621 matching lines @@
     if (native_array_element_type.is_interface_type and
         native_array_element_type.name != 'Dictionary'):
         this_cpp_type = None
         expression_format = 'toMemberNativeArray<{native_array_element_type}>({v8_value}, {index}, {isolate}, exceptionState)'
     else:
         this_cpp_type = native_array_element_type.cpp_type
         if native_array_element_type.is_dictionary or native_array_element_type.is_union_type:
             vector_type = 'HeapVector'
         else:
             vector_type = 'Vector'
-        expression_format = 'toImplArray<%s<{cpp_type}>>({v8_value}, {index}, {isolate}, exceptionState)' % vector_type
+        if native_array_element_type.is_primitive_type:

[haraken, 2017/04/03 04:29:02]

Why do we need to limit this to primitive types?

[bashi, 2017/04/03 04:42:34]

Just to minimize the impact of changes to see if tests are passed. I'm planning
to remove this check :)

[bashi, 2017/04/03 05:06:25]

Looks like we need to improve coverage of native_value_traits_type_name() and
IDLTypes.h to remove this check (e.g. we need to support enums and any). Let me
address it in follow-up CLs.

[Raphael Kubo da Costa (rakuco), 2017/04/06 13:22:36]

(belatedly adding myself to the party)

Are you actively working on this at the moment? I'm also doing some changes to
this part of the code and stumbled upon the problems with enums & friends, but
if you already have a patch we can avoid crossing our streams.

+            value_type = native_value_traits_type_name(native_array_element_type)
+            expression_format = ('toImplArray<%s<{cpp_type}>, %s>'
+                                 '({v8_value}, {index}, {isolate}, '
+                                 'exceptionState)' % (vector_type, value_type))
+        else:
+            expression_format = ('toImplArray<%s<{cpp_type}>>'
+                                 '({v8_value}, {index}, {isolate}, '
+                                 'exceptionState)' % vector_type)
+
     expression = expression_format.format(native_array_element_type=native_array_element_type.name, cpp_type=this_cpp_type,
                                           index=index, v8_value=v8_value, isolate=isolate)
     return expression
 
 
 # FIXME: this function should be refactored, as this takes too many flags.
 def v8_value_to_local_cpp_value(idl_type, extended_attributes, v8_value, variable_name, index=None, declare_variable=True,
                                 isolate='info.GetIsolate()', bailout_return_value=None, use_exception_state=False):
     """Returns an expression that converts a V8 value to a C++ value and stores it as a local value."""
 
@@ skipping 406 matching lines @@
 IdlTypeBase.is_explicit_nullable = property(is_explicit_nullable)
 
 
 def includes_nullable_type_union(idl_type):
     # http://heycam.github.io/webidl/#dfn-includes-a-nullable-type
     return idl_type.number_of_nullable_member_types == 1
 
 IdlTypeBase.includes_nullable_type = False
 IdlNullableType.includes_nullable_type = True
 IdlUnionType.includes_nullable_type = property(includes_nullable_type_union)