- From: Martin Thomson <martin.thomson@gmail.com>
- Date: Tue, 22 Apr 2014 09:51:03 -0700
- To: Amos Jeffries <squid3@treenet.co.nz>
- Cc: HTTP Working Group <ietf-http-wg@w3.org>
On 22 April 2014 08:39, Amos Jeffries <squid3@treenet.co.nz> wrote: >> I can understand if there are END_SEGMENT flags in place that this >> complicates things. But completely-full frames with END_SEGMENT >> should be a pathological case. > > My expectation is that we can cut these frames into two and place the > END_SEGMENT only on the second frame. Exactly. But if you have a 16383 byte frame with END_SEGMENT, then your choices for length on the wire are ==16391 or >=16399. There's no way to put 16392 octets on the wire. I don't think that this is a major issue, since you can be consistent and have both a 16376 byte limit and a new frame for every END_SEGMENT byte. Then the only leakage occurs when END_SEGMENT is used selectively, based on whether there is content that is sensitive/secret/etc...
Received on Tuesday, 22 April 2014 16:51:34 UTC