On Tue, Feb 10, 2009 at 10:05 AM, Evan Martin <[email protected]> wrote: > It's passed into ViewMsg_New, which calls it a "parent window". > Tracing the calls from there, it eventually goes into > RenderWidget::CompleteInit, which stuffs it into host_window_. That's > used in a few places (it appears to be used as a window id, see recent > comments from agl on that) and eventually makes its way into webkit as > a platformWindow.
That was my following of it, yes. The GetPluginNativeView seems like it may be misnamed/overused. > It seems ResourceMessageFilter doesn't have context other than the > arguments supplied by the caller for answering calls like > OnGetWindowRect? But we shouldn't be trusting HWND supplied by the > renderer, so it seems bogus. Right, from the commit log: This highlights a semi-major TODO for the porting effort at some point. We need to abstract the NativeViewIds from the NativeViews as we cannot trust pointer values from the renderers. The conversion code is in base/gfx/native_widget_types.h and currently just casts between them (which is correct for Windows, where both typedefs are HWNDs). One, maybe interresting idea is that we could generate a random secret key in the browser and HMAC sign the pointer values. AGL --~--~---------~--~----~------------~-------~--~----~ Chromium Developers mailing list: [email protected] View archives, change email options, or unsubscribe: http://groups.google.com/group/chromium-dev -~----------~----~----~----~------~----~------~--~---
