Re: Frame Length Restrictions

On Thu, Apr 17, 2014 at 10:05 AM, Martin Thomson
<martin.thomson@gmail.com>wrote:

> On 16 April 2014 17:34, Jeff Pinner <jpinner@twitter.com> wrote:
> > With the addition of padding to the framing layer, I believe it is
> > preferable to implement the frame length requirement at the HTTP layer to
> > allow intermediaries to pad frames without running into frame length
> > restrictions.
>
> The natural follow-on question is: if we allow 64K frames, why not let
> a header block fill the entire thing?  It's not like the smaller frame
> size has a flow control advantage.  The only advantage is in ensuring
> that padding space is available.  But limiting the length to 32K would
> work adequately for that too.
>

The header length limit was meant to force implementations to implement and
test continuation frames and these become more likely with a 16K limit than
a 32K limit.

Personally I feel that this is unnecessary, and that the same could be
achieved with a comprehensive test suite rather than with a MUST in the
protocol, but that wasn't the consensus at the San Francisco interop
meeting.