[geos-devel]createEmptyGeometry()

[strk at refractions.net]

On Fri, Mar 31, 2006 at 01:20:51AM +0200, Mateusz £oskot wrote:
> Martin Davis wrote:
> > I thought about this, but on the principal of Occam's Razor decided 
> > just to use an empty GC.  This has the advantage that there are fewer
> >  cases to handle for both developers and clients.
> 
> That's clear for me.
> 
> strk:
> The only thing I'd do is to review current impl. of GEOS and see if
> there are any possible runtime bugs.

Code is FULL of assumption about the fact that geometries 
are not EMTPY... just took a look at LineString.cpp, and
the CoordinateSequence pointer is dereferenced always w/out
checks.
Note that the same happens with JTS. Example:

  public Coordinate[] getCoordinates() {
    return points.toCoordinateArray();
 }

'points' is not checked for being NULL.
In other places it is.

Now, what does this comport in Java I don't know,
but doesn't look correct to me.

BTW, still we're all learning something, what
would happen by dereferencing a NULL shared_ptr<> ?

Example:
	boost::shared_ptr<Geometry> g;
	g->isEmpty();

Is an exception thrown ?

--strk;
> 
> strk:
> Another thing I'd consider is to check if we could check this behaviour
> to operate with types more strictly. May be it would help to get rid of
> dynamic polymorphism overhead.
> 
> 
> Cheers
> -- 
> Mateusz £oskot
> http://mateusz.loskot.net
> _______________________________________________
> geos-devel mailing list
> geos-devel at geos.refractions.net
> http://geos.refractions.net/mailman/listinfo/geos-devel

-- 

 /"\    ASCII Ribbon Campaign
 \ /    Respect for low technology.
  X     Keep e-mail messages readable by any computer system.
 / \    Keep it ASCII.