Broken chunk encoding on PDF Answered
I am a developer for Sophos (http://sophos.com), and I'm investigating an issue reported by a mutual customer regarding PDF files downloaded from your site.
Our Mac anti-virus product includes a web filtering component. This will scan downloaded files and other content before the browser sees it, but in order to do this it must cache the content until the file download is finished. The pcap file provided to us by the customer suggests that your chunk encoding implementation is missing the '0\r\n\r\n' termination element, and so can never be regarded as complete.
Also, the pcap capture indicates that your chunk encoding implementation puts each byte of the PDF file into its own chunk. While this is technically a correct usage of chunk encoding, it results in a six-fold increase in download size. I'd speculate that the single-byte chunk size is simply another symptom of the same issue which omits the terminator.
I'd be very happy to speak to one of your developers to discuss these issues further.