SDO_UTIL WKT functions appear to choke on multipoints

Question

Hi folks,
Ugh. Not having fun with this. It appears that on both 10.2.0.4 and 11.1.0.6 that the WKT parser doesn't do multipoints correctly.
Could someone familar with WKT check these results and tell me if I am missing something critical?
This works just fine:
<strong>SELECT SDO_UTIL.APPEND(</strong>
<strong>SDO_GEOMETRY('POINT(-87.9033279296583 42.580894084168456)',8265),</strong>
<strong>SDO_GEOMETRY('POINT(-87.89577482907342 42.60414701132971)',8265)</strong>
<strong>) FROM dual;</strong>
producing
SDO_GEOMETRY
2005,
8265,
NULL,
SDO_ELEM_INFO_ARRAY
1,
1,
1,
3,
1,
1
SDO_ORDINATE_ARRAY
-87.9033279296583,
42.5808940841685,
-87.8957748290734,
42.6041470113297
But this
<strong>SELECT</strong>
<strong>SDO_GEOMETRY('MULTIPOINT((-87.9033279296583 42.580894084168456),(-87.89577482907342 42.60414701132971))',8265)</strong>
<strong>FROM dual;
</strong>produces a screwy SDO_ELEM_INFO_ARRAY indicating only a single element and I don't think that 2 is ever legit for a point.
SDO_GEOMETRY
2005,
8265,
NULL,
SDO_ELEM_INFO_ARRAY
1,
1,
2
SDO_ORDINATE_ARRAY
-87.90332792965830321918474510312080383301,
42.58089408416845600413580541498959064484,
-87.89577482907341732243367005139589309692,
42.60414701132970805019795079715549945831
Reversing around the situation, its just as bad
<strong>SELECT SDO_UTIL.TO_WKTGEOMETRY(
(SELECT SDO_GEOMETRY
2005,
8265,
NULL,
SDO_ELEM_INFO_ARRAY
1,
1,
1,
3,
1,
1
SDO_ORDINATE_ARRAY
-87.9033279296583,
42.5808940841685,
-87.8957748290734,
42.6041470113297
) FROM dual)
) FROM dual</strong>
produces a completely whacked
MULTIPOINT ((-87.9033279296583 42.5808940841685))
Am I missing something here? The WKT isn't rocket science so I think I have all the formats correct.
Thanks,
Paul

Answer 1

Paul
1. the second example is correct and seen as a point cluster.
2. seems that the wkt parser only recognizes these point clusters rather then multipoints (possible defect).
There is a slight distinction between a Multi-Point and a Pont cluster:
General:
In Oracle Spatial a Multi...geometry has for every element a triplet in the SDO_ELEM_info_array. Also for a Mulitpoint geometry.
INSERT INTO t1 (i, d, g)
VALUES (
17,
'Multipoint',
sdo_geometry (2005, null, null, sdo_elem_info_array (1,1,1, 3,1,1, 5,1,1),
sdo_ordinate_array (65,5, 70,7, 75,5))
However, there is also 'Point cluster' definition, which just has 1 sdo_elem_info_array triplet (1, 1, n) with n the number of elements.
INSERT INTO t1 (i, d, g)
VALUES (
16,
'Point cluster',
sdo_geometry (2005, null, null, sdo_elem_info_array (1,1,3),
sdo_ordinate_array (50,5, 55,7, 60,5))
In the Pro Oracle Spatial book, the last notation is recomended for a multipoint (performance + storage). It is seen as a single elements with N points, as opposed to an array of points.
If you use in your case:
SELECT SDO_UTIL.TO_WKTGEOMETRY(
SDO_GEOMETRY
2005,
8265,
NULL,
SDO_ELEM_INFO_ARRAY
1,
1,
2
SDO_ORDINATE_ARRAY
-87.9033279296583,
42.5808940841685,
-87.8957748290734,
42.6041470113297
) FROM dual
you will get MULTIPOINT ((-87.9033279296583 42.5808940841685), (-87.8957748290734 42.6041470113297))
In the other case, the WKT parser will read the first triplet and will evaluate the 3rd element of that triplet as a "Point cluster" with 1 point, and will use in second step only parse the first two ordinates.
I would raise this as a possible defect (not reading the full Multipoint) or consider working with the Point Clusters if feasible.
Note: read this post also on issues with the Point Clusters and the number of points: Re: Run-time error using SDO_GEOMETRY constructor with WKT
Hoping this could help.
Luc
Edited by: lucvanlinden on Dec 29, 2008 8:00 PM