Re: [Xen-devel] [PATCH v4] libxl: introduce LIBXL_DOMAIN_TYPE_INVALID

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Fri, 2012-05-25 at 14:10 +0100, Dario Faggioli wrote:
> On Fri, 2012-05-25 at 12:03 +0100, Ian Jackson wrote:
> > Ian Campbell writes ("Re: [PATCH v4] libxl: introduce 
> > LIBXL_DOMAIN_TYPE_INVALID"):
> > > so having arranged to call that function at the right time we can assume
> > > that type is a sensible value, and indeed setdefault makes this the
> > > case.
> > 
> > Right.
> > 
> Ok.
> 
> > The other situation where we can get _INVALID is if libxl__domain_type
> > fails, which it can do.
> > 
> > I think this should be handled by having places which call
> > libxl__domain_type abandon operation and return an error if the
> > libxl__domain_type fails.
> > 
> > If this is done, then general variables, parameters, etc. within libxl
> > which are supposed to contain a libxl_domain_type will never contain
> > _INVALID.
> > 
> I like this. I'll chase each call to that function and have the calle
> failing if a DOMAIN_TYPE_INVALID is returned. Then, if I go this way,
> can I also nuke both the 'case DOMAIN_TYPE_INVALID' _and_ the default
> clauses from everywhere? I seem to think I could...

iff the compiler is smart enough to realise that in the type == INVALID
case you have returned already before reaching the switch statement,
otherwise you will need to have "case INVALID: abort()".

> 
> Thanks and Regards,
> Daio
> 



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel