(no subject)

Ewald Salcher ([email protected])
Wed, 12 Apr 95 14:23:32 PDT

JK,

>
> Are all standard types in IDL, standard types in MRL too? If not,
> they will have to be, since the application mrl will have to use it. Thanks.
>

This is the MM_Any like it appears in IDL now:
typedef struct mmany {
MM_Type_Code type;
double d_value;
MM_Object o_value;
MM_Collection c_value;
MM_Expression e_value;
} MM_Any;

The MM_Any (and the MRL) now cover
- the most important primitive IDL types
- the type MM_Object
- the type MM_Collection
- the type MM_Expression (which for now is a CORBA::String)

------------------------------------------------------------

(i) It would be easy to integrate all primitive types:
- all numerical types and char-types are mapped to d_value
- string is already there - it is e_value
(ii) I see also no difficulty for enums - their values map to d_value
(iii) I'm not sure about the strategy for complex types
(structure, union, array, sequence). As I see it there is no way
to bring a structure into MM_Any (equivalent to void* in C++).

Would you realize a structure in the Application Model as a
MM_Object with MM_Attributes representing the members of the
structure? And accordingly the other complex types? So, this
would not go into MM_Any and the MRL.

-------------------------------------------------
Ewald Salcher USC/Information Sciences Institute
4676 Admiralty Way, Marina del Rey, CA 90292
Phone: (310) 822-1511 250 Fax: (310) 823-6714