Re: Application "core protocol" BOF/WG idea

On Fri, 29 Jan 1999, Eliot Lear wrote:
> There are two reasons to do what you suggest, only one of which will
> fly.
> [1]	To provide guidance on not having to repeat the same mistakes made
>       by others.  This is a good thing.

I agree this is a good thing, but I also suspect that by itself it lacks
sufficient benefit to motivate people to do the hard work.  Anyone could
have documented the mistakes of the past at any time.  The closest anyone
came to doing so that I'm aware of was my internet draft:
But I found myself lacking sufficient motivation to continue that hard
work.  It cost a lot of time and had no visible benefit.

> [2]	To provide mechanisms that keep us from existing using
> 	protocols inappropriately.

Given the way you've worded this I'd have to agree.  I don't think the
IETF can or should forbid people from abusing existing protocols.  That's
not something a concensus organization should do.  But we can develop a
simple alternative to abusing existing protocols which addresses most of
motivations for such abuse.  I consider that the vital motivation for this

> I think [1] can be accomplished with a BCP, stating those BCPs (make a
> stronger statement than Informational). 

While I wouldn't oppose it being a BCP, I really don't think that's
necessary.  If we document past mistakes anywhere, we can always point to
it and say "you're repeating this mistake."

> Further, it's not clear that developing an APPLCORE protocol is
> reasonable to attempt, since it will force you into too many LCD
> situations with lots of options.  This is something that we in the IETF
> have struggled with in the past.

I believe it's possible to make reasonable decisions to avoid too many LCD
options.  But this is an interesting point which I will keep in mind.

> "We have seen ISO and they are us."

Funny, I see APPLCORE as an attempt to counter-act one ISOism that's
creeping into the IETF.  ISO tends to require the use of lots of
overcomplex and often unnecessary protocol and encoding layers.  I see
APPLCORE as creating a simpler alternative to some really large and
complex protocol layers (with lots of unnecessary complexity) people are
insisting on using. 

		- Chris

Received on Friday, 29 January 1999 16:06:54 UTC