Re: Re: SRTM Eurasia

I believe the vpf code is David's so perhaps he should review it for
correctness before we commit it to the repository?

Curt.


Alex Romosan writes:
> "Curtis L. Olson" <[email protected]> writes:
> 
> > For what it's worth, I have fetched all of the SRTM Eurasia data,
> > chopped it, and now I am working on the terrafit stage.  There is a
> > lot of data here so it could take a couple days.
> 
> last night i generated the e020n40 chunk and you can get it from
> http://caliban.lbl.gov/fgfs/e020n40/. this is rather detailed as i ran
> terrafit with a min number of nodes set to 250, max number of nodes to
> 3000, and the max error set to 2. i had problems generating the Lakes
> data from the vmap data.
> 
> (running tgvpf --chunk=e020n40 --work-dir=Lakes --material=Lake /home/romosan/scenery/vmaplv0 eurnasia hydro inwatera). 
> 
> it looks like the program gets into an infinite loop while findind
> edges for a contour (inserting the same edge over and over until the
> program runs out of memory). i am not quite sure i understand the code
> but the following patch solved the problem for me:
> 
> RCS file: /var/cvs/TerraGear-0.0/TerraGear/src/Lib/vpf/contour.cxx,v
> retrieving revision 1.1
> diff -u -r1.1 contour.cxx
> --- contour.cxx 18 Sep 2001 21:17:48 -0000      1.1
> +++ contour.cxx 17 Dec 2003 20:40:49 -0000
> @@ -90,18 +91,32 @@
>        previous_edge = current_edge;
>        if (edg.getValue(row, "right_face")
>           .getCrossRef().current_tile_key == _polygon_id) {
> -       info.isLR = true;
> -       current_edge = edg.getValue(row, "right_edge")
> -         .getCrossRef().current_tile_key;
> +        info.isLR = true;
> +        current_edge = edg.getValue(row, "right_edge")
> +          .getCrossRef().current_tile_key;
> +        if (edg.getValue(row, "left_face")
> +            .getCrossRef().current_tile_key == _polygon_id)
> +          std::cout << "right face also left face" << std::endl;
>        } else if (edg.getValue(row, "left_face")
> -                .getCrossRef().current_tile_key == _polygon_id) {
> -       info.isLR = false;
> -       current_edge = edg.getValue(row, "left_edge")
> -         .getCrossRef().current_tile_key;
> +          .getCrossRef().current_tile_key == _polygon_id) {
> +        info.isLR = false;
> +        current_edge = edg.getValue(row, "left_edge")
> +          .getCrossRef().current_tile_key;
>        } else {
>         throw VpfException("edge does not belong to face");
>        }
>  
> +      for ( vector<line_info>::reverse_iterator it = _lines->rbegin();
> +            it != _lines->rend(); it++ )
> +      {
> +        if ( it->id == info.id )
> +        {
> +          std::cout << "eeek!!! infinite loop..." << std::endl;          
> +          _nPoints -= info.size;
> +          return *_lines;
> +        }
> +      }
> +
>        _lines->push_back(info);
>      } while (current_edge != _start_edge);
>    }
> 
> the reason i go back through the vector instead of just checking
> against the last inserted edge is that there is one instance where the
> edges repeat as a pair. trying to understand when this infinite loop
> happens i realized that every time the infinite loop happens the edge
> happens to be both right (which we check for first) and left. i am not
> sure what this means though. i tried reversing the handedness of the
> edge when this happens, but it didn't work reliably. obviously
> something else is happening here.
> 
> and while i am submitting patches, this is another one i applied:
> 
> RCS file: /var/cvs/TerraGear-0.0/TerraGear/src/Lib/vpf/tile.cxx,v
> retrieving revision 1.2
> diff -u -r1.2 tile.cxx
> --- tile.cxx    7 Oct 2002 15:36:26 -0000       1.2
> +++ tile.cxx    17 Dec 2003 20:45:54 -0000
> @@ -12,11 +12,13 @@
>  VpfTile::VpfTile (VpfTableManager &tableManager,
>                   const string &path, int tile_id)
>    : VpfComponent(tableManager, path),
> -    _tile_id(tile_id), // FIXME: kludge
> -    _face_id(-1),
> +    _tile_id(tile_id),
>      _aft(0),
>      _fbr(0)
>  {
> +  const VpfTable &aft = getAFT();
> +  int row = aft.findMatch("id", _tile_id);
> +  _face_id = aft.getValue(row, "fac_id").getInt();
>  }
>  
>  VpfTile::VpfTile (const VpfTile &tile)
> @@ -55,12 +57,6 @@
>    VpfRectangle rect;
>    const VpfTable &fbr = getFBR();
>  
> -  if (_face_id == -1) {
> -    const VpfTable &aft = getAFT();
> -    int row = aft.findMatch("id", _tile_id);
> -    _face_id = aft.getValue(row, "fac_id").getInt();
> -  }
> -
>    int row = fbr.findMatch("id", _face_id);
>    rect.minX = fbr.getValue(row, "xmin").getReal();
>    rect.minY = fbr.getValue(row, "ymin").getReal();
> 
> basically it moves the initialization of _face_id to the constructor
> (where it belongs) and thus removes the ugly kludge.
> 
> --alex--
> 
> -- 
> | I believe the moment is at hand when, by a paranoiac and active |
> |  advance of the mind, it will be possible (simultaneously with  |
> |  automatism and other passive states) to systematize confusion  |
> |  and thus to help to discredit completely the world of reality. |
> 
> _______________________________________________
> Terragear-devel mailing list
> [email protected]
> http://mail.flightgear.org/mailman/listinfo/terragear-devel

-- 
Curtis Olson   HumanFIRST Program               FlightGear Project
Twin Cities    curt 'at' me.umn.edu             curt 'at' flightgear.org
Minnesota      http://www.flightgear.org/~curt  http://www.flightgear.org