RE: DR308: stability

Mark, I like this suggestion. It would get us out of the position of having
ill-defined and untestable requirements. The alternative is to actually
specify measurable attributes for req'ts such as "stability."

Bill

> -----Original Message-----
> From: Mark Nottingham [mailto:mnot@akamai.com]
> Sent: Sunday, November 19, 2000 11:49 PM
> To: XML Distributed Applications List
> Subject: Re: DR308: stability
> 
> 
> 
> One thing that might be a good idea is to migrate "feel-good" 
> requirements
> like this to a "Guiding Principles" section, rather than 
> making them full
> requirements. 
> 
> 
> On Sun, Nov 19, 2000 at 03:56:52PM -0500, David Ezell wrote:
> > On Tue, 14 Nov 2000 16:30:55 -0800 Mark Nottingham wrote:
> > >Stability is a laudable goal, but does it really mean 
> anything in the
> > >context of requirements for the specification? Every 
> protocol designer wants
> > >a stable specification, but it's not always possible to 
> determine this
> > >before it's in use for quite some time.
> > 
> > Actually, I believe you are somewhat correct; however, you 
> could substitute
> > the word "simplicity" for "stability" in the above remark 
> and it would
> > be equally correct.  
> > 
> > It might be the view of some people that including sections 
> on "simplicity"
> > and "stability" is somewhat obtuse; who wouldn't want these 
> in a design?
> > 
> > That said, calling out both "simplicity" and "stability" is 
> proving useful
> > I think.  WRT "stability", it's important to make clear 
> that XP must be much
> > more stable than any given XP application; the requirements 
> document will
> > provide "litmus tests" as we proceed, and I can easily 
> imagine voting against
> > a requirement which, all things being equal, makes XP less stable.
> > 
> > Best regards,
> > -David  
> 
> -- 
> Mark Nottingham, Research Scientist
> Akamai Technologies (San Mateo, CA)
>

Received on Monday, 20 November 2000 13:15:42 UTC