On 9/15/21 20:39, Dmitry Timoshkov wrote:
Michael Stefaniuc mstefani@winehq.org wrote:
@@ -57,19 +57,17 @@ static inline struct cache *impl_from_IAssemblyCache(IAssemblyCache *iface) static HRESULT WINAPI cache_QueryInterface( IAssemblyCache *iface, REFIID riid,
- void **obj )
- void **ret_iface ) {
- struct cache *cache = impl_from_IAssemblyCache(iface);
- TRACE("%p, %s, %p\n", cache, debugstr_guid(riid), obj);
- TRACE("%p, %s, %p\n", iface, debugstr_guid(riid), ret_iface);
- *obj = NULL;
*ret_iface = NULL;
if (IsEqualIID(riid, &IID_IUnknown) || IsEqualIID(riid, &IID_IAssemblyCache)) { IAssemblyCache_AddRef( iface );
*obj = cache;
*ret_iface = iface; return S_OK; }
Is that really necessary to rename obj to ret_iface? That doesn't really make the change clearer IMO, and makes the patch larger.
*obj = iface;
That looks just wrong to me... but I spend probably too much qwality time with COM in Wine.
sxs was one of the earlier COM cleanups so predating some of the scope creep that happened during that effort. As part of the scope creep Jacek and Alexandre agreed on replacing obj with ret_iface for the 3rd parameter of QI. As I already touch two lines with "obj" in them I'm applying that change too.
Also this change is trivial to validate as the exact same object files are generated when compiling with -g0.
Reference: Sadly we lost the COM Cleanup wiki page and https://wiki.winehq.org/Wine_Developer%27s_Guide/COM_in_Wine is a hodgepodge between obj and ret_iface.
bye michael