Opened 13 years ago
Closed 13 years ago
#458 closed defect (invalid)
Polygonizer robustness failure
Reported by: | strk | Owned by: | |
---|---|---|---|
Priority: | major | Milestone: | 3.3.1 |
Component: | Default | Version: | 3.3.0 |
Severity: | Unassigned | Keywords: | |
Cc: |
Description (last modified by )
A simple multiline containing 3 linestrings of 2,5 and 6 vertices, all sharing the same endpoints (2 nodes in total in the topology) fail to be polygonized in floating precision, while successfully polygonize as 2 rings with a precision of 1e-11.
See http://trac.osgeo.org/postgis/ticket/1115 for original report.
In the geometry with 5 vertices, distance between first and second vertex and between forth and fifth vertex is 2.3283064365387e-10.
The polygonizer returns _no_ rings. Dropping one of the close-by vertices (not on the topology node) makes the polygonizer return _1_ ring. Dropping both makes it succeed in returning _2_ rings as it should.
Simplified MULTILINESTRING input follows:
010500000003000000010200000002000000555C8F42354A3A41A4703D6A1E445241CCCCCC0C474A3A415C8FC2851E445241010200000005000000555C8F42354A3A41A4703D6A1E445241545C8F42354A3A41A4703D6A1E445241409B06043D4A3A415860C1CC1C445241CDCCCC0C474A3A415C8FC2851E445241CCCCCC0C474A3A415C8FC2851E445241010200000006000000CCCCCC0C474A3A415C8FC2851E445241C4F5289C4B4A3A415C8FC2C52B445241B533D71C964A3A418EAD7CBD1A4452412085EB91284A3A417A14AE2719445241285C8F42354A3A41A4703D6A1E445241555C8F42354A3A41A4703D6A1E445241
Change History (3)
comment:1 by , 13 years ago
Summary: | Polygonizer failure → Polygonizer robustness failure |
---|
comment:2 by , 13 years ago
Description: | modified (diff) |
---|
comment:3 by , 13 years ago
Resolution: | → invalid |
---|---|
Status: | new → closed |
It seems that the input is still not properly noded. In particular there's a linear intersection between the line with 5 vertices and the one with 6. Passing the geometry under unaryunion turn it to a 4 lines geometry, and that one successfully polygonizes.