[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Need one more draft
On Jan 31, 2007, at 2:38 AM, Jan Algermissen wrote:
On Wednesday, January 31, 2007, at 05:26AM, "Tim Bray"
<Tim.Bray@xxxxxxx> wrote:
As far as I know, this is the first
application-level protocol ever to be designed based on this
assumption, and thus there is benefit in making it painfully explicit
right up front.
Hmm.....I find this a bit troubling. HTTP was the first
**application-level protocol** ever to be designed based on this
assumption, right?
I wouldn't say HTTP was the first, but it was definitely designed that
way intentionally and protocols layered on top of it should be as well.
Independent components cannot be free to evolve independently if they
are making implementation assumptions about each other's purpose.
Bit-for-bit storage is one of those assumptions.
Putting it another way, this is exactly equivalent to people who say
that every Internet protocol should use TCP, including the protocols
that are transmitting lossy live video feeds. We should not make
irrational design comments about the horror of losing packets when
it is known quite well that a live video protocol is designed as a
system of trade-offs, with one of them being that lost packets are
much less harmful than waiting for a retransmission.
Paragraphs like those proposed simply reflect a lack of understanding
of the communication goals for which the protocol was designed. The
response should be to describe those goals in the abstract and
introduction so that the AD understands them, not simply toe the
line because it is politically expedient. I don't want ADs to make
technical decisions for working groups.
....Roy