You can clone with
HTTPS or Subversion.
It'd be nice to know how many bytes the various http parsing functions consume, so that what's happening at that layer can be accurately mapped to what's actually in the packets.
Yeah, we need to somehow expose the difference between packets and chunks in http_trace. Right now, they are sort of magically often the same, but not always.