Re: h2 frame layout

On 30 August 2014 09:04, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote:

> --------
> In message <25E37A5F-6120-4D84-A1EC-2830C0B41537@gbiv.com>, "Roy T.
> Fielding" w
> rites:
>
> >  0                   1                   2                   3
> >  0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
> > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > |                   Stream Identifier (32)                      |
> > +---------------+-----------------------------------------------+
> > | Frame Type(8) |   Payload Length (24)                         |
> > +=+=============================================================+
> > |                   Frame Payload (0...)                      ...
> > +---------------------------------------------------------------+
>
> >
> >Rationale for B:
> >
> >  The current usage of frame flags in spec-14 is frightening.
> >  In all cases, they can and should be separate frame types.
>
> I'm 100% behind with Roy here.
>
>
​This lines up pretty closely with my suggestion in the
END_STREAM/CONTINUATION thread [1]. I'm +1 for removing flags from the
general frame format header.

I'm much less keen on reducing the size of frame_type field, because that
would limit my ability to play with my toys^H^H^H^Hextensions[2][3].


[1]: http://lists.w3.org/Archives/Public/ietf-http-wg/2014JulSep/1313.html​
​[2]: http://tools.ietf.org/html/draft-kerwin-http2-encoded-data-03
[3]: ​http://tools.ietf.org/html/draft-kerwin-http2-segments-01


-- 
  Matthew Kerwin
  http://matthew.kerwin.net.au/