Re: [PATCH 3/4] oleacc: Add ObjectFromLresult implementation

[Michael Stefaniuc]

On 04/25/2014 02:18 PM, Dmitry Timoshkov wrote:
> Piotr Caban  wrote:
> > > > On 04/25/14 13:34, Qian Hong wrote:
> > > > > - atom = result;
> > > > > - if(atom != result)
> > > > > -
> > > > >    return E_FAIL;
> > > > >
> > > > > I can't understand these few lines, are they typo or did I miss anything?
> > > >
> > > > result and atom has different size. This code checks if LRESULT
> > > > describes atom or error.
> > >
> > > GlobalGetAtomNameW should handle this just fine, 'atom' is a misleading
> > > intermediate variable IMO.
> >
> > No, it's not handling it.
>
> Why? Are there any casts in the way that break this?
> > It can happen that cast from an error code is 
> > a valid atom.
>
> In that case a simple test for HIWORD(result) would look less convoluted.

In principle that is correct but you cannot use HIWORD on a 64 bit
integer. You'll need something like IS_INTATOM from kernel32 / ntdll.

bye
    michael