RE: Flushing the Message Queue

["Hollenbeck, Scott" <shollenbeck@verisign.com>]

>-----Original Message-----
>From: OneHandyMan@excite.com [mailto:OneHandyMan@excite.com]
>Sent: Tuesday, June 19, 2001 2:04 PM
>To: ietf-provreg@cafax.se
>Subject: Re: Flushing the Message Queue


>Please forgive me but I am starting to get confused here. I was under
notion
>that we are working protocol design and not on system design or
>implementation. 
>
>The protocol should mainly focuses on the message syntax and its context. 
>Issues like dealing with duration of message remain on queue and how to
deal
>with the messages after a certain period of time or should we pass them via
>ftp / e-mail or idl implementation, all are system design and
implementation
>issue.

A protocol specification should provide enough information to build
complete, interoperable implementations.  The document describes a queuing
mechanism, but it currently says nothing about a very real operational issue
that every implementer is going to have to face.  I asked the question
because I would much prefer to make a statement in the document that
acknowledges the issue so that implementers know that we didn't fail to
recognize it, and that it's something they need to be aware of.

I do agree that the mechanics of how to deal with it do not belong in the
protocol document.

<Scott/>