On Sat, Jul 16, 2016 at 10:07:42AM +0200, Christian Mueller wrote: > Hi Frank > > First I want to thank you for the contribution. Time support is quite a > challenge because it should be available to all supported databases > including Oracle GeoRaster, Oracle native, DB2,.....
I agree it *should* work on all of those, but unless someone actually has access to all of them, I don't think adding support for everything at the same time is very realistic. As I see it, the best we can do is try to ensure the shared code doesn't make assumptions that won't hold on systems that aren't updated right away. I've started on pgraster because that's what we want to use here, and I can definitely work on generic postgresql and generic jdbc. Mysql might be possible (I can definitely install it, but I don't have much mysql experience at all), but I don't have access to oracle or db2 servers. > Additionally the time support should work with GeoServer as described here > http://docs.geoserver.org/latest/en/user/services/wms/time.html#specifying-a-time Yes. That works right now (it's what I use for testing), except possibly for periodicity (I'll check and fix if needed). There is an open question though: what's supposed to be returned if more than one raster tile from the database matches the request for a given location? Latest data for each point? Only points for the latest timestamp? Some sort of average? I assume "latest data for each point" makes the most sense, but it's also the hardest to do. I know my code right now doesn't really handle this at all (you'll get data, but which particular bit isn't deterministic). Only points for the latest timestamp seems fairly easy to handle, but latest data for each point is tricky. There are several ways I've looked at: * No multithreaded decoding or composing, so an ORDER BY time on the database query will enforce "latest" or "earliest". This won't work for averages, and dropping multithreading is a rather high price to pay. * Have the database query do the filtering. For pgraster, as far as I can see this implies having either consistent tiling for all data (i.e. tiles for different times have exactly the same grid) or having the query (slowly...) retile everything. For systems without in-database raster systems I don't see a way to do it. > The next point is that some database engines (DB2,Oracle) have a built in > time support which is slightly different to the time support provided by > geotools. Do you have some pointers to documentation? Frank > Nevertheless I want to hear your opinion. > > Cheers > Christian > > -- Frank Gevaerts frank.gevae...@fks.be fks bvba - Formal and Knowledge Systems http://www.fks.be/ Schampbergstraat 32 Tel: ++32-(0)11-21 49 11 B-3511 KURINGEN-HASSELT Fax: ++32-(0)11-22 04 19 ------------------------------------------------------------------------------ What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic patterns at an interface-level. Reveals which users, apps, and protocols are consuming the most bandwidth. Provides multi-vendor support for NetFlow, J-Flow, sFlow and other flows. Make informed decisions using capacity planning reports.http://sdm.link/zohodev2dev _______________________________________________ GeoTools-Devel mailing list GeoTools-Devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-devel
