Re: [Swig-user] %newobject and %typemap(newfree)

[Petr N. <Pet...@am...>]

Thanks a lot, I am checking your recommendation.

Interesting note: To use that, I must artificially make A::~A public, even=
 though it is intentionally hidden (private) in the original class design.=
 Without a public destructor, delete_A is not generated and %unrefobject=
 has nothing to apply to. I guess it's a feature... although I can imagine=
 that an explicit %unrefobject should trigger delete_A generation.

BTW, I still think I did exactly what 10.5.9 asks me to do. But I see what=
 you mean: The "newfree" is supposed to free the pointer that the called=
 function supposedly allocated by the invoking function; the %newobject=
 does not belong there - it applies only when the returned pointer is=
 wrapped as a pointer (and it tells target language whether it owns the=
 object the pointer points to) and not if the returned value is translated=
 to a native type. Ergo, I think 10.5.9 has two problems - it is confusing=
 about purpose of newfree and incorrect about usage of %newobject.

Regards,

--
Petr Novotn=FD
Prague Development Center, AMDOCS
+420266773112, +420608711517
"First Law of Bad Management: If something isn't working, do more of it." -=
 Tom DeMarco=0D

> -----Original Message-----
> From: John Lenz [mailto:jl...@ma...]=0D
>=0D
> On 10/23/06 10:45, Petr Novotny wrote:
> > Hi,
> >=0D
[snip]
> > After reading the documentation, I thought that I am supposed to
> > inform swig by %newobject that B::test creates a new object, and by
> > %typemap(newfree) A* how A should be dispatched of.=0D
>=0D
> No, the newfree typemap is for freeing temporary memory created during
> the typemap, that is used to build the real return value.  You should
> instead try using
> %unrefobject A "A::recycle($this)"
[huge snip]
>=0D
> John
>=0D
>=0D

This message and the information contained herein is proprietary and=
 confidential and subject to the Amdocs policy statement,
you may review at http://www.amdocs.com/email_disclaimer.asp