#POSTGIS IRC Log - 2009-05-13

For logs after Feb 3, 2007, all times are GMT-8. Prior logs are GMT-9.
Back to Logs
02:31:00 sigq: geosfeed: Ticket #249 ( defect updated ): GEOS, Shapely, DLLs, and memory <http://trac.osgeo.org/geos/ticket/249#comment:4>
03:22:45 CIA-34: warmerdam * r2507 /trunk/capi/ ( geos_c.cpp geos_ts_c.cpp ): forward GEOSFree to GEOSFree_r, use std:: prefix on free( ) ( #249 )
03:25:51 CIA-34: warmerdam * r2508 /trunk/capi/geos_c.h.in: refer to GEOSFree( ) instead of free( )
03:27:16 sigq: geosfeed: Ticket #249 ( defect updated ): GEOS, Shapely, DLLs, and memory <http://trac.osgeo.org/geos/ticket/249#comment:5>
03:30:57 sigq: geosfeed: Ticket #244 ( defect updated ): Buffer fails to buffer elements of multpolygon <http://trac.osgeo.org/geos/ticket/244#comment:13>
04:30:57 CIA-34: warmerdam * r2509 /branches/3.1/ ( 4 files in 2 dirs ): added GEOSFree( ) and GEOSFree_r( ) ( #249 )
04:32:41 sigq: geosfeed: Ticket #249 ( defect closed ): GEOS, Shapely, DLLs, and memory <http://trac.osgeo.org/geos/ticket/249#comment:6>
07:43:08 CIA-34: mloskot * r4080 /spike/wktraster/scripts/gdal2wktraster.py:
07:43:08 CIA-34: [wktraster] Added new command line option -O, --overviews for overview tables
07:43:08 CIA-34: creation and population being under development. Reorganized utility functions.
07:43:08 CIA-34: Eventually, code of the script should be modularised into a couple of classes.
10:30:13 sjzzalx: hello all. I'm having a problem building postgis on Ubuntu ... this seems to be a common problem, but thing doesn't find the postgres header files, though they seem to exist in the right location. Any known fix for this?
11:57:50 jlivni: sjzzalx: can you be more specific with what you're up to? works fine for me
11:58:13 jlivni: following http://os.umbrellaconsulting.com/wiki/UbuntuBaseStackSource for example works
11:58:14 sigq: Title: UbuntuBaseStackSource - trac_os - Trac ( at os.umbrellaconsulting.com )
13:47:28 nhv: anyone ever try to have multiple level of detail of line objects in a table ?
13:52:26 pramsey: nhv: in one table? no
13:53:22 pramsey: the google encoding of lines has some odd thoughts along those lines which are going to make writing a serializer for that format tricky http://code.google.com/apis/maps/documentation/polylineutility.html
13:53:23 sigq: Title: Interactive Polyline Encoder Utility - Google Maps API - Google Code ( at code.google.com )
13:53:40 nhv: I guess having a separate table for each LOD makes more sense
13:55:23 nhv: another newbie question is it absurd to be building tables with 100s of lines where at full detail each line has 100s of K points ?
13:57:52 nhv: ya LOD lines are a tricky concept
13:58:16 pramsey: yes, absurd
13:58:35 pramsey: very expensive to pull each one out, and probably you only want a small portion of it for each calculation
13:58:50 pramsey: better to chop them into < 200 vertices
13:59:05 pramsey: so that each sub-component fits in a page
13:59:21 nhv: hmm ...
14:00:27 nhv: that is sort of what I do with the raw data except I have 10 minutes of data approx 2500 points per segment
14:00:30 pramsey: the actual length you can fit will vary depending on how much other stuff you have in your tuple and how many dimensions are in your line
14:00:59 nhv: that works relatively well in my ascii flat file db
14:02:15 nhv: maybe I can add a segment ID to each record an build an index on those and accomplish sort of the same
14:02:56 pramsey: track_id + sequence_id
14:03:05 pramsey: or somesuch : )
14:03:12 nhv: track_id + track_sement_id
14:03:17 pramsey: aye
14:04:01 nhv: can I expect this to scale to many millions of points
14:04:36 nhv: I am storing scientific data that is being collected at 5 times a second
14:05:35 nhv: and trying to improve on a custom ascii heirarchial db
14:05:50 nhv: I really just need better spatial partitioning
14:07:12 nhv: actual data is measured in TBs but metadata which I want DB access to is in 10s of GBs
14:07:19 nhv: think image database
14:08:58 nhv: anyway thanks
14:09:19 nhv: I am going to try adding a second key to the tracks
14:36:54 pramsey: nhv: if you cut things up under the page size, you should find things scale quite well. spatial partitioning is something I need to investigate...