Re: Content-Transfer-Encoding: packets for HTTP

John Ellson (ellson@hotair.att.com)
Tue, 27 Sep 94 21:13:57 EDT


> From: "Daniel W. Connolly" <hotsand!hal.com!connolly>
> To: Multiple recipients of list <www-talk@www0.cern.ch>

[...]

> The default content-transfer-encoding inside an aggregate/* is packet. So for
> example:
>
> 0200 document follows
> Content-Type: aggregate/mixed
> Content-Transfer-Encoding: binary
>
> Content-Type: text/html
>
> 4000
> ...4000 bytes of stuff...
> 1745
> ...1745 bytes of styff...
> 0
> Content-Type: image/gif
>
> 4000
> ...4000 bytes of stuff...
> 1745
> ...1745 bytes of styff...
> 0
> .
>
>
> Does this seem like a reasonable proposal?
>
> Dan

I like this proposal. It requires finite buffers (4000 bytes) which
will improved performance of piped applications, and it supports multiparts.

Just a question, can the receiver of http assume a reliable transport
mechanism, without errors, or should there be some kind of CRC or
similar confidence check on the length headers to assure that the
receiver doesn't get misaligned?

...well two questions... Are you proposing a standardized maximumum size
on the packets, or could this be negotiated in an Accept header?
(4K sounds like a reasonable fixed max to me .... anything but 53bytes ;-)

John Ellson
AT&T Bell Labs