Re: [3/5] msxml3/xsl: Basic put_input() method for IXSLProcessor

[Jacek Caban]

Hi Nikolay,

On 2/14/11 9:04 PM, Nikolay Sivov wrote:
> Basic put_input() method for IXSLProcessor

+    IXMLDOMNode *old_input = This->input;
+    HRESULT hr;

-    FIXME("(%p): stub\n", This);
-    return E_NOTIMPL;
+    TRACE("(%p)->(type=%d)\n", This, V_VT(&input));
+
+    /* try IXMLDOMNode directly first */
+    if (V_VT(&input) == VT_UNKNOWN)
+        hr = IUnknown_QueryInterface(V_UNKNOWN(&input),&IID_IXMLDOMNode, (void**)&This->input);
+    else if (V_VT(&input) == VT_DISPATCH)
+        hr = IDispatch_QueryInterface(V_DISPATCH(&input),&IID_IXMLDOMNode, (void**)&This->input);
+
+    if (hr != S_OK)
+    {
+        IXMLDOMDocument *doc;
+
+        hr = DOMDocument_create(&CLSID_DOMDocument, NULL, (void**)&doc);
+        if (hr == S_OK)
+        {
+            hr = IXMLDOMDocument_QueryInterface(doc,&IID_IXMLDOMNode, (void**)&This->input);
+            IXMLDOMDocument_Release(doc);
+        }
+    }
+
+    if (hr == S_OK&&  old_input) IXMLDOMNode_Release(old_input);

Failure of QueryInterface call will set This->input to NULL (you probably want it not changed and even if not, that's a leak). It's probably better idea to have local pointer to the new input instead of old one and change it once you know the function will succeed.

Jacek

[3/5] msxml3/xsl: Basic put_input() method for IXSLProcessor

[Jacek Caban]

On 2/15/11 4:53 PM, Jacek Caban wrote:
> Hi Nikolay,
>
> On 2/14/11 9:04 PM, Nikolay Sivov wrote:
> > Basic put_input() method for IXSLProcessor

+    TRACE("(%p)->(type=%d)\n", This, V_VT(&input));

I missed it in my previous mail, but we have debugstr_variant for this in msxml3 now.

Jacek

[3/5] msxml3/xsl: Basic put_input() method for IXSLProcessor

[Nikolay Sivov]

On 2/15/2011 18:53, Jacek Caban wrote:
> Hi Nikolay,
>
> On 2/14/11 9:04 PM, Nikolay Sivov wrote:
> > Basic put_input() method for IXSLProcessor
>
> +    IXMLDOMNode *old_input = This->input;
> +    HRESULT hr;
>
> -    FIXME("(%p): stub\n", This);
> -    return E_NOTIMPL;
> +    TRACE("(%p)->(type=%d)\n", This, V_VT(&input));
> +
> +    /* try IXMLDOMNode directly first */
> +    if (V_VT(&input) == VT_UNKNOWN)
> +        hr = 
> IUnknown_QueryInterface(V_UNKNOWN(&input),&IID_IXMLDOMNode, 
> (void**)&This->input);
> +    else if (V_VT(&input) == VT_DISPATCH)
> +        hr = 
> IDispatch_QueryInterface(V_DISPATCH(&input),&IID_IXMLDOMNode, 
> (void**)&This->input);
> +
> +    if (hr != S_OK)
> +    {
> +        IXMLDOMDocument *doc;
> +
> +        hr = DOMDocument_create(&CLSID_DOMDocument, NULL, (void**)&doc);
> +        if (hr == S_OK)
> +        {
> +            hr = IXMLDOMDocument_QueryInterface(doc,&IID_IXMLDOMNode, 
> (void**)&This->input);
> +            IXMLDOMDocument_Release(doc);
> +        }
> +    }
> +
> +    if (hr == S_OK&&  old_input) IXMLDOMNode_Release(old_input);
>
> Failure of QueryInterface call will set This->input to NULL (you 
> probably want it not changed and even if not, that's a leak). It's 
> probably better idea to have local pointer to the new input instead of 
> old one and change it once you know the function will succeed.
> Right, another thing is that I don't load a document from input variant.
>
> Jacek