This bug is a holder for any nits that we'd like to clean up in ipc/mscom, but were inconvenient to do so at the time they were surfaced. At the time I had a massive patch queue and it was just too disruptive to make those changes using hg histedit. One of them is renaming utils.* to Utils.* The other is deciding where to put DynamicallyLinkedFunctionPtr -- jimm suggested nsWindowsHelpers. Other nits and suggestions are welcome.

Other suggestions: * Ability to filter out QI logging (jimm); * Remove 'Unique' from pointer names in ipc/mscom/Ptr.h. tbsaunde points out that even though we're (ab)using UniquePtr to wrap COM objects with untouchable refcounts, they're still not "Unique", so it's confusing.

Trevor pointed out that: mParentProxy.reset(parentProxy.Release()); is a bit messy. It would be nice if we could to an assignment operator here that would permit moving from a COMPtrHolder to a COMPtrHolder::COMPtrType

Also: IAccessible* rawIntercepted = nullptr; GetPointer(&rawIntercepted); IAccessibleHolder::COMPtrType iaIntercepted(rawIntercepted); return IAccessibleHolder(Move(iaIntercepted)); Is too verbose. Let's clean up COMPtrHolder construction so that we can make this cleaner.

- Encapsulation for VARIANTS so that we can easily instantiate one for CHILDID_SELF with less verbosity

(In reply to Aaron Klotz [:aklotz] from comment #4) > - Encapsulation for VARIANTS so that we can easily instantiate one for > CHILDID_SELF with less verbosity This one isn't really necessary: since CHILDID_SELF == 0, one can just do something like: const VARIANT kVarChildIdSelf = {VT_I4};

Not working on this right now.