On Dec 10, 2008, at 5:38 PM, Henrik Nordstrom wrote: > ons 2008-12-10 klockan 17:25 -0800 skrev Roy T. Fielding: > >> I say "no" -- close the loophole and remove any notion that multipart >> delimits message framing. > > Unfortunately that's not entirely an option. We need to have parsers > still understand this delimiting of 206 responses as they are in > widespread use and specified so in 2616 (and 2068). They aren't used that way. I don't know of any implementation that uses the multipart delimiter to find the end of body (they use CL, chunked, or a closed connection). > But I am entirely fine with denying servers from using this method of > delimiting and instead wrap multipart/byteranges in chunked encoding, > and keeping the parser requirements only for compatibility with > earlier > HTTP/1.1 specifiations. I am entirely fine with removing multiple byteranges as a feature, since I regard it as a DoS problem, but that's another matter. ....RoyReceived on Thursday, 11 December 2008 02:13:16 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:34 GMT