linestring instead of point? about mobilitydb

[estebanzimanyi]

Dear Wendell

Many thanks for your interest in MobilityDB. If I understand well you problem, what you need is to compute the temporal intersection of a moving point and a geometry. You can do this in MobilityDB, as shown in the following example

select tintersects(tgeompoint '[point(0 0)@2000-01-01, point(3 3)@2000-01-04]', geometry 'Polygon((1 1,1 2,2 2,2 1,1 1))');
-- "{[f@2000-01-01 00:00:00+01, t@2000-01-02 00:00:00+01, t@2000-01-03 00:00:00+01], (f@2000-01-03 00:00:00+01, f@2000-01-04 00:00:00+01]}"

Please notice that a moving point may enter and leave a geometry multiple times as shown in the following example

select tintersects(tgeompoint '[point(0 2)@2000-01-01, point(6 2)@2000-01-11]', geometry 'Polygon((1 0,1 1,2 3,3 1,4 3, 5 1,5 0,1 0))');
"{[f@2000-01-01 00:00:00+01, t@2000-01-03 12:00:00+01, t@2000-01-05 04:00:00+01],
(f@2000-01-05 04:00:00+01, t@2000-01-06 20:00:00+01, t@2000-01-08 12:00:00+01],
(f@2000-01-08 12:00:00+01, f@2000-01-11 00:00:00+01]}"

If you want to obtain the first and last instant at which at moving point intersects a geometry you can obtain them as follows

select period(atValue(tintersects(tgeompoint '[point(0 2)@2000-01-01, point(6 2)@2000-01-11]', geometry 'Polygon((1 0,1 1,2 3,3 1,4 3, 5 1,5 0,1 0))'), true))
"[2000-01-03 12:00:00+01, 2000-01-08 12:00:00+01]"

Regarding your question about using generic geometries instead of points in MobilityDB, the current version does not support them since all temporal operations (e.g. temporal intersection) are much more complex for moving geometries than for moving points. Nevertheless, we are already exploring the issue in order to implement OGC's Moving Features standard https://www.opengeospatial.org/standards/movingfeatures
Therefore, yes, we will be able to support moving geometries (which are non deforming) in a future release of MobilityDB.

Regards

Esteban

[wendellwt]

Thank you for the reply. I'll make use of MobilityDB with points that have timestamps. That will be much better than PostGIS with the LinestringM.

Thanks again,
Wendell