Is the definition of security_ace_object in librpc/idl/security.idl correct?

[Richard Sharpe]

Hi folks,

Perhaps I am mistaken, but my reading of the following from
librpc/idl/security.idl:

        typedef struct {
                security_ace_object_flags flags;
                [switch_is(flags & SEC_ACE_OBJECT_TYPE_PRESENT)]
security_ace_object_type type;
                [switch_is(flags &
SEC_ACE_INHERITED_OBJECT_TYPE_PRESENT)]
security_ace_object_inherited_type inherited_type;
        } security_ace_object;

suggests that the secuity_ace_object_type and
security_ace_object_inherited_type GUIDS will only be
marshalled/unmarshalled if the appropriate bits are set.

However, my reading of [MS-DTYP].PDF, section 2.4.4.3 suggest that
those fields are present on the wire regardless of the bit values and
the flags field only serves to tell us whether those fields are valid.

Can anyone confirm?

I ask in the context of bug 9821 because there seems to be evidence
(not yet confirmed by a torture test) that Windows does not like it
when either of those fields is not physically there in the structure.

-- 
Regards,
Richard Sharpe
(何以解憂？唯有杜康。--曹操)