Analyze not creating spatial statistics on LineString Z geometry

[rightway]

I’ve loaded a line shapefile containing height information into PostGIS using OGR2OGR. The shapefile has loaded OK and the geometry all seems to be fine in PostGIS, showing as a LineStringZ geometry.

When I run st_estimated_extent on the table it fails with the error message ‘ERROR: stats for “testlayer.shape” do not exist. When I run an analyze on the table it says "NOTICE: no non-null/empty features, unable to compute statistics"

Running st_isvalidreason(shape) on the table says that all the geometry is valid.

I haven't had the chance to test any other types of 'Z' or 'M' geometry but I assume they behave the same way.

We are running on: PostGIS 2.1.5 64 bit Postgres 9.3.5 64 bit Windows 2012

[pramsey]

Could you attach the file to the ticket? I need to recreate the problem to fix it. Thanks!

[rightway]

I've uploaded a sample file. I notice that ogr2ogr defaults to loading the shapefile as a linestringZ, while the postgis shapefile loader loads it in as MultiLinestringM, but either way extents are not calculated.

[pramsey]

The problem is that the third dimensions is garbage

308,289680.5366 693147.6176 -1.79769313486232e+308

Every vertex seems to have -1.79769313486232e+308 as the M value. (Since ogr2ogr doesn't understand M, it has to interpret it as Z)

[rightway]

Thanks, finally managed to create a valid linestringZ dataset and verified that st_estimatedextent does work with it.

I guess that my only comment is that if the geometry isn't valid enough to analyze then it would be helpful if the st_isvalid and st_isvalidreason commands picked up this error. Either that or get the analyze to ignore dodgy Z and M values and still work out the extents which is only a box2d anyway.