On Thu, 2007-11-01 at 08:23 -0700, Steve Meier wrote:
> I fail to see how this prevents having multiple views for a model.

> The viewer provides the drawing function and graphics context. The
> library just uses what the viewer provides. If the viewer wants one
> window to display one representation then the viewer calls draw page
> with a particular gc that gets passed along to each of the pages sub
> objects.

It implies that all multiple views must co-operate via a common
interface. Lets say for example that my "viewer" might not be graphical?
It might be a feed-forward to gnucap, it might be an attribute editing
window (NB: Just an example, I don't intend to re-write gattrib).

In these cases, I don't care about repainting objects behind a changed /
damaged object, I just care about the changed object. In the gschem
case, I might need to repaint overlapping objects. This then requires me
to have redraw logic in gschem _anyway_ to find the overlaps, and
request they be redrawn.

I'm of the opinion libgeda should be an file IO, data-structure and EDA
database library. Its remit should not include "driving" a GUI.

> But the nice thing is ... is that the viewer can be very ignorant of the
> internals of the library structures.

That is the PCB approach, and I don't think its right for gschem. If
gschem doesn't know about libgeda's API, then what good is it. I think
we should have libgeda enabling EDA operations, not defining the
operating policy for any app which wants to use it.

Peter C.



_______________________________________________
geda-dev mailing list
[email protected]
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

Reply via email to