[Clam-devel] Copy & Paste processings on canvas patch

[Pau Arumí Albó]

El dv 13 de 06 de 2008 a les 17:01 +0200, en/na David García Garzón va
escriure:
> On Divendres 13 Juny 2008, Pau Arumí Albó wrote:
> > El dv 13 de 06 de 2008 a les 15:13 +0200, en/na David García Garzón va
> >
> > escriure:
> > > > >  Also that
> > > > > way of doing will ask for every processing even if you don't have
> > > > > geometry for it (transition period, see previous comment) so you
> > > > > don't have a way to know whether (0,0,0,0) was a real geometry or
> > > > > just it hadnt geometry at all.
> > > >
> > > > Easy, let (*,*,0,0) be a non-existing geometry (we don't want invisible
> > > > boxes).
> > >
> > > Yes, Pau but you should compare the size with zero in order to use the
> > > one in the pos file. For me is even clearer to iterate along a map and
> > > set just the ones that are present. At that point it is just a matter of
> > > taste.
> > >
> > > if (posfile exist)
> > > 	load pos file
> > > dict = network.GetAndClearGeometries()
> > > for key, geometry in dict
> > > 	boxes[key].setGeometry(geometry)
> > >
> > > c++ code would be not that clear but i like this procedure.
> >
> > I also like this scheme, but the point is "do we want to couple the
> > Canvas with two new Network types?"
> > Specially when we are in the process of abstracting the Canvas so it can
> > be used with an abstract Model interface (that is, in non NetworkEditor
> > contexts)
> >
> > What about that? Do you buy it?
> >
> > for key in Canvas:
> >         int x,y, w,h
> >         bool exists
> >
> >         network.getGeometry(x,y,w,h,exists)
> >         if not exists: continue
> >         boxes[key].setGeometry(x,y,w,h)
> >
> > P
> 
> This is already network coupled code. I mean, having to load or storing a clam 
> network is something in ClamNetworkCanvas, not the abstract NetworkCanvas.


Ok, good to know. Then my suggestion evaporates :-)

P