#POSTGIS IRC Log - 2009-07-01

For logs after Feb 3, 2007, all times are GMT-8. Prior logs are GMT-9.
Back to Logs
00:35:10 mleslie: M-P: PostGIS has been rid of the heavy weight datatype for some time, so there's no longer a strong need to call the datatype light weight. It is however, the light weight datatype of old.
00:48:29 M-P: mlechner: yes i can see that all i got now is just geometry datatype but it makes my DB bloat after millions of points.
00:49:01 M-P: now i got about 40 million points and it's wasting gigabytes of space and index is size of 2 giga
00:49:07 M-P: no so elegant solution
00:52:00 mlechner: M-P: talking to mleslie not me, eh?
00:53:49 M-P: oh, sorry
00:54:19 M-P: mlechner: feel free to participate :D
01:09:45 mlechner: M-P: I've got only 1200 points to handle - no problems with large indexes and really boring for my postgresql. So I think i won't be much help.
03:48:33 CIA-2: robe * r4234 /trunk/doc/using_postgis_dataman.xml: amend documentation on curved support to include new compound curve in curve polygon support.
03:49:20 CIA-2: robe * r4235 /branches/1.4/doc/using_postgis_dataman.xml: amend documentation on curved support to include new compound curve in curve polygon support.
04:26:10 CIA-2: robe * r4236 /trunk/doc/reference.xml: arg should be text not string for st_buffer
07:20:22 clj2289: pramsey, would you have any interest in being on FLOSS weekly? I listen to the show and I think it would be very interesting to hear you being interviewed... http://twit.tv/FLOSS
07:20:23 sigq: Title: The TWiT Netcast Network with Leo Laporte ( at twit.tv )
07:20:56 pramsey: I have a never-ending appetite for self-promotion, clj2289 .
07:21:48 clj2289: haha, would it be cool with you then if I emailed them and attached your name and this ( postgis ) project?
07:22:01 pramsey: sure, that would be wonderful
07:22:18 clj2289: cool, what email should I use to CC you?
07:22:36 pramsey: pramsey@cleverelephant.ca
07:23:40 clj2289: sounds good, and thanks
07:37:20 CIA-2: pramsey * r4237 /branches/1.4/postgis/ ( lwgeom_accum.c lwgeom_geos.c ): Fix bad memory access in aggregates on nulls ( #210 ), Mark Cave-Ayland.
07:37:26 CIA-2: pramsey * r4238 /trunk/postgis/ ( lwgeom_accum.c lwgeom_geos.c ): Fix bad memory access in aggregates on nulls ( #210 ), Mark Cave-Ayland.
08:07:39 darkblue_B: hiya
08:07:56 darkblue_B: so.. GEOS 3.1.0 C API 1.5.0
08:08:08 darkblue_B: is the new and spiffy setup yes?
08:08:20 darkblue_B: PostGIS svn trunk
08:11:57 CIA-2: pramsey * r4239 /branches/1.4/ChangeLog: Updated ChangeLog for 1.4.0rc1
08:12:20 pramsey: GEOS 3.1.1
08:12:31 darkblue_B: ah so
08:15:27 CIA-2: pramsey * r4240 /tags/1.4.0rc1/: Tag 1.4.0rc1
08:17:33 CIA-2: pramsey * r4241 /tags/1.4.0rc1/ ( NEWS README.postgis Version.config ): Update version number to rc1 in tag.
08:28:40 pimpaa: hello pramsey
08:28:47 pimpaa: i was looking at some tickets in trac
08:29:07 pimpaa: see ticket #180, i think i can help with it
08:30:21 pramsey: OK. It's sort of a memory placeholder for me, I didn't really spec it out.
08:30:59 pimpaa: pramsey, its just to create a example of "audit" tables/triggers and functions right?
08:31:03 pimpaa: and write about it?
08:31:50 pramsey: more of a working infrastructure...
08:31:52 pramsey: select postgis_enable_history( 'thistable' )
08:32:00 pramsey: creates 'thistable_history'
08:32:08 pimpaa: ok, got it
08:32:23 pramsey: and hooks up rules to 'thistable' to manage 'thistable_history'
08:32:25 pimpaa: and set up the trigger
08:32:29 pimpaa: sure
08:32:34 pimpaa: i can give it a try!
08:32:45 : * pramsey recommends learning the power of RULES
08:32:52 pramsey: if you don't know them already
08:33:12 pramsey: if you can fit a problem into the RULE space, you'll get more efficient performance
08:33:29 pramsey: since triggers just get re-written into rules, and since triggers usually involve a PL while rules do not
08:33:43 pimpaa: oh right
08:33:48 pramsey: they can't do as much as triggers, but for what they can do, they are very elegant
08:34:20 pimpaa: sure, ill see what i can do about it
08:34:28 pimpaa: how much time do i have?
08:34:34 pramsey: ages
08:34:38 pimpaa: lol
08:34:40 pramsey: 1.5 isn't coming for a while
08:34:51 pimpaa: ill see what i can do with that 'short' time
08:35:36 pimpaa: its a nice feature, and since its just sql i think i can help
08:51:45 CIA-2: pramsey * r4242 /branches/1.4/CREDITS: Bring this up to date.
08:51:54 CIA-2: pramsey * r4243 /trunk/CREDITS: Bring this up to date.
08:55:02 pimpaa: pramsey, do i need to create an account on postgis trac?
08:55:15 pramsey: yes, to do anything other than read
08:55:39 pramsey: http://www.osgeo.org/osgeo_userid
08:55:41 sigq: Title: OSGeo Services Userid | OSGeo.org ( at www.osgeo.org )
09:02:05 darkblue_B: wow - trunk build failed
09:02:25 darkblue_B: for me.. I'll look for this RC1 package I guess
09:03:00 darkblue_B: no rule to make target 'FuzzyPointLocater.cpp'
09:03:54 darkblue_B: ( it was building fine until I .configured again )
09:09:04 darkblue_B: make clean && make
09:09:09 darkblue_B: trying again
09:19:25 darkblue_B: rev 2607... same result
09:28:40 hobu: darkblue_B: ./autogen.sh is probably needed with a configure
09:38:31 darkblue_B: oh
09:38:34 darkblue_B: ok, trying
09:39:39 darkblue_B: doh - forgot to mention, its trunk GEOS, to get to the 3.1.1
09:39:48 darkblue_B: still, trying autogen
09:46:41 darkblue_B: hobu: GEOS trunk still fails to build.. ah well
09:46:55 darkblue_B: I suppose I need to find this 3.1.1 prebuilt then
09:47:54 darkblue_B: why is this one file different?
09:47:55 darkblue_B: weird
12:43:44 Schuyler: hallo
12:43:50 Schuyler: there isn't a GEOS specific IRC channel, is there?
12:48:05 pimpaa: i dont think so
12:59:41 Schuyler: is there any documentation for the GEOS C API aside from the geos_c.h header file?
13:24:37 : * Schuyler waves at pramsey
13:24:44 pramsey: hi Schuyler
13:27:21 pramsey: no, the header file is it
13:27:30 pramsey: darkblue_B: re-checkout the source and build
13:28:00 pramsey: I have no idea what the problem is, but at some point that FuzzyPointLocater.cpp thing cropped up, and a clean checkout seems to be required to lose it
13:28:08 pramsey: ( hip hip hooray for autotools )
13:31:05 Schuyler: pramsey: is there any documentation for the GEOS C API aside from the geos_c.h header file?
13:31:21 pramsey: no, the header file is it ChanServ
13:31:24 pramsey: Schuyler:
13:34:31 Schuyler: pramsey: ?
13:34:45 pramsey: just the header file
13:35:12 pramsey: are you seeing this, Schuyler or is Colloquy eating your irc?
13:49:01 Schuyler: pramsey: I saw your last two msgs
13:50:58 pramsey: hey TheSteve0
13:52:21 Schuyler: what's the status of the PreparedGeometry code in GEOS?
13:52:41 pramsey: complete, rocking, in use
13:56:15 Schuyler: ok. is Geometry::setUserData( ) not exposed in the C API for some reason?
13:56:37 Schuyler: also, will PreparedGeometry optimize relate( ) calls?
13:58:27 pramsey: the only reason things are exposed in the CAPI is because someone said "hey, I need to use that".
13:58:39 Schuyler: ah
13:58:55 Schuyler: the Doxygen docs are quite emphatic about using the C API over the C++ API
13:59:01 Schuyler: is ABI between versions the only reason?
13:59:18 pramsey: and Prepared geometry will not optimize relate calls, because relate calls naturally have to compute the complete DE9IM which is what prepared geometry is built to avoid doing.
13:59:30 pramsey: preparedgeometry is to get fast answers to true/false predicates
13:59:38 Schuyler: ahh I see
14:00:01 pramsey: yeah, the only reason. strk had a real fear-of-ABI-change period in his life
14:00:23 pramsey: and ended up trying to scare everyone off of the C++ API, since it was practically impossible to keep the ABI stable.
14:00:41 Schuyler: FrankW seems to have had that too. correct me if I'm mistaken, but it just means that anything compiled against GEOS might have to be recompiled if GEOS is upgraded?
14:00:44 pramsey: But, frankly, if you don't mind re-compiling your software when a new release comes out...
14:01:06 pramsey: correct. the code API is pretty stable, so a re-compile is the only pain, as a general rule.
14:01:15 Schuyler: well I'd like to make life easier on my client in this case, but there's a cost/benefit analysis involved
14:01:37 Schuyler: thanks very much for the info, pramsey
14:01:40 pramsey: np
14:44:44 Schuyler: so #include <geos.h> throws a bunch of deprecated warnings in 3.0.0
14:44:53 Schuyler: what *should* I be including to get the Geometry class?
14:45:01 Schuyler: the warning doesn't say.
14:47:37 pramsey: Sorry, Schuyler, don't know!
16:27:50 CIA-2: mleslie * r4244 /branches/1.4/ ( 3 files in 2 dirs ): Fixed the inappropriate minpoint checking of circular strings within curvepoly, as reported in ticket 212, comment... 7ish.
19:07:57 hobu: Schuyler: you should be using the C API instead of the C++ one