On 4 July 2014 07:34, Martin Thomson <martin.thomson@gmail.com> wrote: > > 2. A mandate to fill all frames carrying header block fragments if > they are followed by a CONTINUATION > > In the normal case, this would mean that an implementation could avoid > even implementing CONTINUATION. > > That is, if we get this right and say that padding can be used to fill > the frame, otherwise we're back in the poop. > > Do you mean "padding can't be used"? Otherwise I can still send an empty HEADERS, padded out to the full frame size, and put all my real bytes in a CONTINUATION. Because I'm a jerk. Conversely is this proposal oppositional to padding? If you want to obscure the length of your header block, and you can fit within a single HEADERS frame, then all well and good; but if you know/suspect you're going to overflow a single frame, you're kinda SOL. If you're going to overflow by a calculable amount, according to the current draft you can defer more bytes to the continuation and pad out the first HEADERS. If you can only continuate when the HEADERS is legitimately full, you can't do anything to obscure the length. Would this proposal reopen the "padding on CONTINUATION" issue? And if it's not considered significant, why is the padding on headers to start with? -- Matthew Kerwin http://matthew.kerwin.net.au/
Received on Thursday, 3 July 2014 23:49:53 UTC