> From: Lars Clausen <lrclause@cs.uiuc.edu>
> Date: 2003/08/13 Wed AM 11:28:59 EDT
>
> On 13 Aug 2003, ashalper@cox.net wrote:
> > This seems related to the current discussion on views. A layer is a
> > piecewise disjoint subset of objects (please correct me if I'm wrong
> > here--I don't use layers that much) in which the object positions are
> > fixed.
>
> That's one way to describe it. Also, there is a z-ordering on the layers.
Please excuse my ignorance, but what's z-ordering?
> > A view (as I am proposing it) would be an arbitrary subset of
> > objects in which the object positions are allowed (but not required) to
> > vary. Effectively an independent diagram with the restriction that the
> > included objects must be a subset of the "base" diagram.
> >
> > I've been thinking, and I was wondering if it would cause anyone pain if
> > we extended the current Dia view function to the description above?
>
> So as you change from view to view, the same object may have different
> positions?
But is not required to have different positions. The default is every object has the same position as it has on the base diagram.
> That sounds complex. It means that an object no longer just
> has one position, like many things assume, but have a number of possible
> positions depending on the view.
It is complex, but it's really the only practical way to decompose complex UML or ER diagrams that represent "spaghetti-code" legacy systems (which wind up looking like, well, spaghetti). I don't think it's too bad really (I say this before I attempt to implement it of course). n subdiagrams of a diagram.
> > It seems like I could do it so that the basic original view capability is
> > preserved. There would probably be some additional interface complexity,
> > but I think the end result would be more useful than current views.
>
> Right now, I'm not sure it's worth the additional effort, it'd be a major
> code change.
OK. I'll twiddle around with it for my own use for now.
Andy