[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH] 2/3 Store the range size when adding a DHCP range



On Wed, Oct 14, 2009 at 02:07:15PM +0100, Daniel P. Berrange wrote:
> On Wed, Oct 14, 2009 at 02:31:29PM +0200, Daniel Veillard wrote:
> > On Wed, Oct 14, 2009 at 12:33:36PM +0100, Daniel P. Berrange wrote:
> > > > > 
> > > > > It would be nice to have this in a callable function too
> > > > > 
> > > > >      int virSocketAddrRange(struct sockaddr_storage *start,
> > > > >                             struct sockaddr_storage *end);
> > > > 
> > > >   Are you supposed to look struct sockaddr_storage ? As posted in my
> > > > last mail this seems a completely opaque structure at least in theory
> > > > and if you want to keep the portability it's supposed to bring.
> > > 
> > > You cast to one of the address specific structs according
> > > to the ss_family field.
> > 
> >   humpf ... okay it has to be cast to be accessed, that's weird,
> > definitely.
> 
> Yeah, ideally it would have been in a union, but POSIX the way it was
> defined when sockets API were first designed, didn't allow a union to
> be retrofitted. 
> 
> We could consider a union ourselves though for our API if it might 
> make it a little clearer, eg
> 
>    typedef union {
>       struct sockaddr_storage stor;
>       struct sockaddr_in inet4;
>       struct sockaddr_in6 inet6;
>    } virSocketAddr;
> 
> to allow more direct access without the casting

 Ah, yes, that look way better that way, and use that for all our
 internal APIs. Fine then, I will look at this,

 Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel veillard com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | virtualization library  http://libvirt.org/


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]