- From: Jim Gettys <jg@freedesktop.org>
- Date: Tue, 14 Jul 2009 07:26:28 -0400
- To: Carsten Bormann <cabo@tzi.org>
- CC: Julian Reschke <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>
Carsten Bormann wrote:
> On Jul 14, 2009, at 12:48, Jim Gettys wrote:
>
>> that it be shown to be significantly more compact than HTTP
>
> It is worth pointing out that saving bytes is one goal; saving code
> complexity is another significant goal. Tight binary coding with few
> options is much easier to parse than a lenient text-based protocol on an
> 8-bit processor with 48 KiB of flash.
>
> That said, whether the result of such an effort still is HTTP is the
> interesting question.
>
Exactly. Though it's not clear that the stated devices are having to
parse the protocol, rather than just a client implementation generating
a small subset of HTTP.
Of course, it also isn't clear how many people still use 8 bit
processors with 48K of flash (which aren't probably running an IPv6
stack, either).
And yes, HTTP parsing is radically more complex and takes a lot more
code than roughly equivalent binary code frameworks. But you don't have
to write them from scratch, either. And for reasons I've never
understood, some folks seem to think writing such binary protocols is
hard. They are just different (though many binary protocols have made
insufficient/incorrect provisions for extensibility). X11, for example,
made some (non-fatal) mistakes in that area.
- Jim
Received on Tuesday, 14 July 2009 11:27:11 UTC