[Date Prev] [Date Next] [Thread Prev] [Thread Next] Indexes: Date | Thread | Author

[XML-SIG] WSDL library ?


Rich Salz <rsalz@zolera.com> writes:    (01)

> Thanks, Martin, for the explanation of how Corba can do doubly-linked
>> lists.  I am glad it's gotten more competent since I looked.  Can it
>> really do full cycles now, where head->next = tail->prev?  Cool.    (02)

Indeed. It was introduced to be able to implement Java RMI on top of
it.    (03)

> I claimed most distributed computing doesn't use the Corba object
>> model. DCOM doesn't, it's aggregation of interfaces.  Java uses single
>> inheritance and interface.  Neither of those can be mapped into a
>> simple inheritance model.    (04)

CORBA's interface inheritance is not single; any interfaces can have
many base interfaces. For valuetypes, single inheritance is only
available for "stateful" valuetypes; there is also the notion of
abstract valuetypes, which do support multiple inheritance. Again,
you can do all of Java RMI on top of CORBA.    (05)

Implementing COM-style interface navigation on top of CORBA is
straight-forward:    (06)

interface IUnknown{
  IUnknown get_interface(in UUID target);
};    (07)

interface ISomething:IUnknown{
  ...
};    (08)

This is exactly what DCOM does.    (09)

> As for being "surprised" or "horrified" that all pointers are chased
>> when you serialize an object.  Hunh?  If I declared an object in Corba
>> or DCE IDL (or MIDL) I would be quite upset if only certain parts of
>> it ended up being remoted.    (010)

AFAIK, in DCE IDL, you have have a void* field in an struct, and then
you use attributes of the field (like unique, size, etc) to indicate
whether this pointer may be shared, and whether the pointer is to
a single object or an array of objects, in which case you need somehow
specify the size of the array. It seems easy to get wrong.    (011)

Regards,
Martin    (012)