#POSTGIS IRC Log - 2008-08-06

For logs after Feb 3, 2007, all times are GMT-8. Prior logs are GMT-9.
Back to Logs
09:16:30 darkblue_B: bitner: one of the postgres people pointed out.. the failure in the import was early on, then the \N become errors.. nothing to do with DB encoding or character set or anything.. you can set a flag to make the import stop on failure
09:16:55 darkblue_B: .. or hit ^c quickly, before the original error msg scrolls out of buffer
14:39:04 mloskot: pramsey: I'm not forcing to release 3.1 now, I was just asking for comments or proposing it
14:39:18 mloskot: simply, I've not known anything about prepared geoms leaking
14:39:21 mloskot: etc.
14:39:25 mloskot: just to explain
14:39:26 pramsey: well, that's my comment, but your reasons for wanting 3.1 are valid
14:39:45 pramsey: alternatively, we could release 3.0.1, even though we have added to the C api
14:39:55 pramsey: since mostly you want to expose the bug fices
14:39:55 mloskot: but I agree that it's a good idea to eliminate this bug for 3.1
14:39:56 pramsey: fixes
14:40:11 mloskot: yes
14:40:28 mloskot: ok, so we need someone brave enough to make a decision, hehe
14:40:32 mloskot: hobu: ? ,- )
14:40:38 pramsey: I can do that, I have no shame
14:40:45 mloskot: fantastic : )
14:40:50 pramsey: Although...
14:40:51 pramsey: erm.
14:40:52 : * mloskot is good i delegating
14:40:57 pramsey: where are the bug fixes?
14:41:02 pramsey: on trunk or 3.0.x?
14:41:07 mloskot: trunk
14:41:08 pramsey: we did actually branch, if I'm not mistaken
14:41:24 mloskot: we are talking about 3.0.1 now, right?
14:41:24 hobu: -1 on release 3.0.1 with API changes
14:41:41 mloskot: hobu: good and safe
14:41:41 hobu: breaks the contract we laid out wrt C API
14:41:42 pramsey: thank god, a -1, now hobu will find the memory leak so we can release 3.1 : )
14:41:46 mloskot: but we still need a compromise
14:41:59 hobu: valgrind FTW?
14:42:19 hobu: developers who add new functionality are required to fix their own leaks :P
14:42:21 pramsey: I don't believe it's been ground, or if the bug has even been isolated to either geos or postgis
14:42:37 pramsey: well, I've been hounding benjubb for a while, but I'm not his boss anymore
14:43:34 mloskot: we have "only" 3 bugs with leak word in their reports
14:43:35 mloskot: http://trac.osgeo.org/geos/search?q=leak&noquickjump=1&ticket=on
14:43:36 sigq: Title: Search Results - GEOS - Trac ( at trac.osgeo.org )
14:43:48 mloskot: is any of them that one
14:43:49 mloskot: ?
14:44:41 mloskot: if I have good bug report + test case I can use to reproduce it, I make some love with Mrs Valgrind :- )
14:46:14 pramsey: the test cases are postgis + geos
14:46:18 pramsey: I don't have a pure geos one...
14:46:38 pramsey: basically though, run a prepared polygon against several 1000 points
14:46:42 pramsey: and see if it leaks
14:46:50 pramsey: if it doesn't, the problem is in postgis instead
14:47:16 mloskot: pramsey: you mean run as just create big poly ?
14:47:23 mloskot: and see if it gets released on "exit"
14:47:54 pramsey: make/acquire a big polygon, prepare it, and run the contains( ) test on it against a largish collection of points both within and outside it
14:48:26 mloskot: got it
14:48:52 mloskot: I can assign my brain time to it for weekend
14:49:38 mloskot: 1 day is nothing in comparison to 12 months of delay to schedule :- )
14:50:43 pramsey: speaking of delay, I got some more provenence review done today : )
14:50:55 pramsey: just putting together the historical commit list now
14:52:18 mloskot: right, this is annoying work
14:52:35 mloskot: btw, do we want to keep maintaining ChangeLog file?
14:52:37 mloskot: manually
14:53:50 pramsey: could do... if not a build target might be nice
14:53:58 pramsey: so it can be built for releases
14:58:10 mloskot: "build target" ?
14:59:09 pramsey: make ChangeLog
15:02:18 mloskot: is there such command?
15:02:31 mloskot: I think it would be much easier to follow GDAL's approach
15:02:41 mloskot: Here it is:
15:02:58 mloskot: 1. Do care of important changes
15:03:07 mloskot: 2. Important changes have ticket submitted first
15:03:18 mloskot: 3. Each ticket is assigned to a milestone
15:03:37 mloskot: 4. Trac does the dirty job to report changelog for each release using milestone property
15:03:59 mloskot: 5. We publish changelog on Wiki, automagically generated + dump to ChangeLog file.
15:04:00 mloskot: EOF
15:05:37 pramsey: I was thinking more of having a ChangeLog file in the distributed source still, but pulled out of SVN log messages, with svn2cl
15:06:26 mloskot: Why not to get it this way?
15:06:26 mloskot: http://trac.osgeo.org/gdal/wiki/Release/1.5.2-News
15:06:27 sigq: Title: Release/1.5.2-News - GDAL - Trac ( at trac.osgeo.org )
15:06:30 mloskot: and just write to file?
15:07:27 pramsey: because it's not comprehensive, it's per-release
15:08:19 mloskot: yes
15:08:26 mloskot: ok, svn2cl works forme
16:42:51 pimpaa: hello all
16:42:54 pimpaa: bitner hi
16:52:56 darkblue_B: pimpaa: hey
16:57:46 pimpaa: hey darkblue_B
16:57:50 pimpaa: how r ya man?