FrameLoader forwarding calls to FrameLoaderClient
In looking at ways to decrease the agony associated with FrameLoader, I've noticed that there are a whole of 1-line functions on FrameLoader that just call a similarly named function on FrameLoaderClient without accessing any member variables other than m_client. On the one hand, removing these forwarding functions could move a lot of noise out of FrameLoader. On the other hand, doing so would add another dereference at the callsite and increase the number of places we have to include FrameLoaderClient.h. I'm just curious whether this was a conscious design decision (especially since there are also a bunch of functions on FrameLoaderClient that are called directly, rather than having a corresponding function on FrameLoader). Does anyone have insight into the history of this abstraction or strong feelings on it? ~Nate
Originally, I think Maciej and I intended the frame loader client to be private to frame loader, and all calls were intended to flow through the frame loader. The same for other clients elsewhere in WebKit. It would be good to be consistent one way or the other, I guess. Not a very momentous decision. -- Darin
participants (2)
-
Darin Adler
-
Nate Chapin