Re: TypeCode operations

[Sam Couter <scouter bigpond net au>]

Michael Meeks <michael@ximian.com> wrote:
> 
> On Sun, 2002-02-24 at 04:49, Sam Couter wrote:
> > I'm specifically worried about the call to
> > ORBit_RootObject_duplicate() to copy the TypeCode for the CORBA_any, and
> > the assumption that the caller owns the returned CORBA_any. If these two
> > things are correct, then I'm confident that the code is right.
> 
> 	They are correct.

Good. That means I've been reading the spec correctly then.

But if I've been reading the spec correctly, then there are ownership
problems with nearly all of the objects returned from the TypeCode
operations. None of them are duplicated or copied, so even though the
spec says that the caller must free the returned storage, doing so would
break things in ORBit.

Is this a real bug, or an error in my perception?

It's easy for me to fix: Add calls to the appropriate duplicate methods
before returning the object.

It would also need to be fixed in ORBit STABLE, since that's where I got
the code from in the first place.

> 	I think we should work on the cpp branch and merge that to HEAD when we
> branch ORBit2 for Gnome 2.0.0, sadly it seems we'll have to wait until
> 2.0.1 or somesuch for your work to get exposed.

That's no real problem. I'm happier to keep my mistakes out of the HEAD
branch anyway. ;)

> 	Woah - yes, lots and lots of regression tests are good in
> test/everything.

I started adding lots of tests yesterday. Tedious work. :/
-- 
Sam "Eddie" Couter  |  mailto:scouter@bigpond.net.au
Debian Developer    |  mailto:eddie@debian.org
                    |  jabber:sam@jabber.topic.com.au
OpenPGP fingerprint:  A46B 9BB5 3148 7BEA 1F05  5BD5 8530 03AE DE89 C75C

PGP signature