Hi Gabriel,
On 23.03.2020 14:53, Gabriel Ivăncescu wrote:
diff --git a/dlls/jscript/jscript.c b/dlls/jscript/jscript.c index 3fcaab0..eb328d2 100644 --- a/dlls/jscript/jscript.c +++ b/dlls/jscript/jscript.c @@ -757,6 +757,9 @@ static HRESULT WINAPI JScript_SetScriptSite(IActiveScript *iface, hres = retrieve_named_item_disp(pass, item); if(FAILED(hres)) return hres; }
/* For some reason, Windows clears the CODEONLY flag */item->flags &= ~SCRIPTITEM_CODEONLY;
This clearly shows that we're still not interpreting those flags correctly and this patch just tries to hide the problem. I think we simply shouldn't depend on the flag but rather on item->disp being set. See the attached patch (on top of your patches). It solves some of test failures that you observed. The remaining problem is that we still store IDispatch for visible items. The attached test shows that we probably shouldn't.
Thanks,
Jacek