[QUANTITY] our fault for carrying on

[Martin Hill]

Thanks Patrick - that at least, as you say, gives some context!

Any serious discussion does belong on this list however.  I understand 
some private discussion does (and should) go on to clear up concepts 
between individuals.  But this list exists so that people can at least 
monitor discussions and cry 'but that won't work for me!' before we 
commit to anything serious.  People can always put a filter on or 
unsubscribe and resubscribe.

I must admit, possibly due to sand in my ears, that I don't remember 
seeing anything near this concrete at the Cambridge or Strasbourg 
interops...

I know also it's a pain coming to a consensus, and I sympathise (after 
all, I'm desperately waiting for consensus so I can code it!).  But the 
whole point is so that people can indeed drag out their own ideas and 
compare and contrast, so we can see if we've missed anything fundamental.

Largely these discussion groups seem to be about concepts and language 
rather than results - but we need these before we can get a consensus 
that can be used to produce results.  It just takes ages...

Cheers,

Martin

Patrick Dowler wrote:

> A couple of weeks ago, I emailed a UML diagram to ~10 people for comment.
> Instead, we got a pretty ferocious email onslaught that eventually spilled 
> over to the mailing list when 1 of those ~10 people unilaterally decided
> to re-start the discussion on the list and thus get rid of the off-line 
> discussion. That's water under the bridge at this point...
> 
> Just so everyone knows what started it up, I include for your perusal that
> original UML diagram - without further comment since those were largely
> ignored the first time around. I will point out the following: 
> 
> - there are no attributes, maning the model is incomplete and a point for
>  disussing the relationship of the various parts... consider anything that is
>  underspecified to be "open to discussion" or "possible to do" if you like
> 
> - color notation: only the pink bits are what I call "quantity"; the rest if 
>   there for context only, which means that some parts are declared   
>   "out-of-scope" by this diagram
> 
> - UML notation: closed triangle-head arrows mean generalisation, filled
>   diamonds mean aggregation, arrows mean association; I have tried to use
>   the least restrictive and most vague UML concepts possible 
> 
> So, I don't really expect or want more discussion on this since the list is 
> now way into details best left for after one agrees on the basics. This is 
> "the basics" as I see it after looking at many ideas. It is largely the same 
> as the Unified Domain Model of Lemson, Dowler, and Banday (ADASS poster)
> simply because that model drew from all the discussions and proposals over the 
> summer. This is definitely NOT the CVO data model; I'm kind of sick of people
> just dragging out their own model in responce to any attempt at consensus.
> This IS a consensus of ideas brought forward from [Cambridge Interop, ADASS].
> 
> And yes - many things are not in there. Everything cannot be in there because 
> many things are logically inconsistent with others. I don't care how many use 
> cases you throw at it and how many requirements you specify, attempts to 
> build on a logically inconsistent model are doomed to failure. 
> 
> OK, that's my rant for this week.
> 
> 
> 
> ------------------------------------------------------------------------
> 

-- 
Software Engineer
AstroGrid @ ROE
Tel: +44 7901 55 24 66
www.astrogrid.org