Ick, ick, ick. We're trying to move towards core_output_filter *not* to have to process everything if the connection is aborted or an error is detected. Right now, we do do that, but that is wrong for a whole host of reasons we're slowly figuring out (see semi-recent thread between Jeff and myself and some others about this).If the connection is aborted after the metadata bucket is pushed into the output filter chain, it would be the core output filter's job to consume and destroy all remaining buckets until it had processed the request-cleanup metadata bucket.
If an error is detected (or the client aborts), we may not necessarily make it to core_output_filter. -- justin
