Re: CONTINUATION Frame Size

On 14 August 2013 07:24, Manger, James H
<James.H.Manger@team.telstra.com> wrote:
> There is one extra complexity: to support single header name/value pairs that can't fit in a frame.

That's the problem that the current scheme chooses to solve, not the
other.  The compress-into-a-frame with separate frames is more
complicated.  (I'll admit, not by much, but then some is always
greater than none.)

Here's my understanding of how we reached the current design:  We
don't want headers to overflow their allocated space.  Nor do we
expect them to for most uses of the protocol.  It's an edge case.  We
acknowledge the need to accommodate these edge cases, but the
complexity overhead any accommodation imposes has to be the absolute
minimum.

Received on Wednesday, 14 August 2013 16:37:37 UTC