[postgis-devel] lwgeom->bbox

[Sandro Santilli]

Still I don't think computation has to be forced, so the idea
of an argument flag for bbox computation (deep or shallow)
seems good to me.

Of course computing intermediate boxes while computing the outermost
wouldn't add much overhead, if not the one required to allocate
the boxes (which wouldn't be there if we keep boxes by value in the
structure).

But on first deserialization we might have outer box w/out computing,
thus saving the time/space required to compute at all levels.

--strk;

On Tue, Nov 01, 2011 at 11:07:32PM +0100, Nicklas Avén wrote:
> Sorry, forget it. 
> I see that was exactly what you meant Paul. You remembered perfectly
> right.
> 
> /Nicklas
> 
> 
> 
> On Tue, 2011-11-01 at 23:01 +0100, Nicklas Avén wrote:
> > Hallo
> > 
> > > Well, for the objects we're not currently already computing it
> > > (points) that's not a big deal, but I can see how subcomponents might
> > > be an issue. But I recall that Nik had to fight with this a bit in his
> > > work since he needed to compute boxes before he could get started with
> > > his distance routines...
> > 
> > Yes, but that was at sub geometry level. So I needed bboxes of every sub
> > geometry in the multi geometry cases. 
> 
> 
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel

-- 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html