!standard B.03        (71)                            05-02-28  AI95-00337/03
!class binding interpretation 03-07-31
!status Amendment 200Y 04-12-02
!status WG9 Approved 06-06-09
!status ARG Approved 8-0-2  04-11-20
!status work item 03-07-31
!status received 03-01-13
!qualifier Clarification
!priority Low
!difficulty Easy
!subject Applicability of C interfacing advice to private types

!summary

The rules of B.3(66-71) apply to private types by applying the appropriate
rule for the full type.

!question

Several of the rules given in the Implementation Advice section of B.3
are framed in terms of specific kinds of types. For example, the rule
for t* parameter passing in B.3(69/1) is described as applying to
parameters of record types. It would seem sensible for the same
advice to pertain to parameters of a private type whose full type
is a record type. Are these rules also intended to apply to private
types whose full types are in the given classes of types? (Yes.)

!recommendation

(See wording.)

!wording

Add after B.3(71):

An Ada parameter of a private type is passed as specified for the full view
of the type.

!discussion

Parameter passing properties are the same for all views (see 6.2). These also
are parameter passing properties, and thus they should be consistent for all
views.

!corrigendum B.03(71)

@dinsa
@xbullet<An Ada parameter of an access-to-subprogram type is passed as a
pointer to a C function whose prototype corresponds to the designated
subprogram's specification.>
@dinst
An Ada parameter of a private type is passed as specified for the full view
of the type.

!ACATS test

An ACATS test could be constructed to test this, but it would be difficult and
somewhat system-dependent.

!appendix

!topic Applicability of C interfacing advice to private types
!reference RM95-B.3(66-71)
!from Gary Dismukes 03-01-13
!discussion

This question arose because of a customer inquiry.

Several of the rules given in the Implementation Advice section of B.3
are framed in terms of specific kinds of types.  For example, the rule
for t* parameter passing in B.3(69/1) is described as applying to
parameters of record types.  It would seem sensible for the same
advice to pertain to parameters of a private type whose full type
is a record type.  Are these rules also intended to apply to private
types whose full types are in the given classes of types?

****************************************************************

From: Pascal Leroy
Sent: Tuesday, January 14, 2003  4:14 AM

AI 117 states that "The convention of the partial view of a private type or
private extension is the convention of the full type." The AI was written is
this fashion because it was argued that the parameter passing mechanism has to
be determined by looking at the completion of the type (see paragraphs 2.a and
2.c of the !discussion section).  I was uncomfortable with this at the time
because it constitutes a breach of privacy, but there was no other reasonable
option.  Anyway now that the camel's nose is in the tent it would seem logical
to say that the rules in B.3(66-71) are applied by looking at the completion of
the type.

****************************************************************

From: Robert Dewar
Sent: Tuesday, January 14, 2003  4:22 AM

I agree. By the way, this is not just theory, we had a customer who
specifically worried that a private record might not be covered by the rules as
stated.

****************************************************************