When to use void[] vs. ubyte[] in D

Is there a general rule as to开发者_StackOverflow when I should use void[] instead of ubyte[]? Is either preferred?

void[] for data that may contain pointers, byte[] for data that's actually just data, like network buffers or files. (I realize Phobos is inconsistent on this)

The reasoning is that the GC scans void[]s for pointers, but not ubyte[]s.

Check out this discussion:

http://www.digitalmars.com/d/archives/digitalmars/D/learn/272.html