Re: [libxml++] Re: different behavior on solaris/x86 and linux/x86 of Document::create_root_node?

[Christophe de VIENNE <cdevienne alphacent com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Le Vendredi 22 Août 2003 15:10, Murray Cumming a écrit :
> On Fri, 2003-08-22 at 14:51, Stefan Seefeld wrote:
> > > I would suggest a little debugging, maybe into libxml too. Some code
> > > somewhere is probably purposefully setting a pointer to 0 for some
> > > reason.
> >
> > my guess would be that there is a problem with the global initializer
> > object not being created. (that problem was already reported once upon
> > a time on a windows platform).
> >
> > Remember that libxml++'s wrapper mechanism is based on a libxml2
> > callback that has to be initialized first. If it isn't, no
> > wrappers will ever be created.
> >
> > The fact that this mechanism is so fragil and not very portable
> > (or so it seems) made me change the approach on my own libxml2
> > wrapper...
>
> _If_ this is the problem, and there is no simple solution, then maybe we
> need an xmlpp::initialize() function instead.

This suggest that with the same compiler a static instance of a class is 
instanciated or not depending on the platform : quite odd.
I would be very surprised if it is the case.


Cheers,

Christophe
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE/RnzIB+sU3TyOQjARAjQ7AJ9rq3n//EGntYi76lSxo5y2D0VD1gCeIgzp
7/YcITMVLu0CuaMeHi74ueQ=
=VzwT
-----END PGP SIGNATURE-----