| 00:01.05 | brlcad | I rarely test from install unless I'm nearing release testing, several extra minutes saved per compile, which adds up fast |
| 00:04.42 | starseeker | nods - I'm the same way |
| 00:05.06 | starseeker | it's a possible difference to check - the cmake build and the autotools build runs are quite different |
| 00:05.56 | brlcad | nods |
| 00:06.45 | brlcad | yeah, I just verified that your halfspace test failed for half on linux too |
| 00:06.56 | brlcad | it's the oed.mged set of tests |
| 00:07.04 | starseeker | hmm - wonder if I just commented it out and moved on... |
| 00:07.19 | brlcad | I was running them manually since they're easier to isolate |
| 00:07.23 | starseeker | some of those were a little tricky to set up, so I may have just assumed I had done it wrong |
| 00:07.30 | starseeker | nods |
| 00:07.37 | brlcad | so it maybe always has failed |
| 00:08.10 | brlcad | thing is, it stops the whole testing progression and stops pretty early on with it trying to move a half |
| 00:08.24 | starseeker | possibly - that has to be over a year since I've touched those, and probably longer, so I don't recall much |
| 00:08.24 | brlcad | with the fix, it gets all the way to the end |
| 00:08.29 | starseeker | sweet! |
| 00:08.39 | brlcad | successfully even |
| 00:08.45 | brlcad | now the log is another matter... |
| 00:09.00 | brlcad | heh, it lists slews of errors and failures, but it's not clear which are intentional and which are not |
| 00:09.12 | brlcad | e.g., saw a "killall*" in there |
| 00:09.35 | brlcad | but then that might have been a prefix or something too |
| 00:09.38 | starseeker | urm... don't recall if I got to the intentional failure checking or not... |
| 00:10.00 | starseeker | mainly recalls a lot of time spent on primitives and quick reference card commands |
| 00:10.52 | starseeker | brlcad: you still build primarily in the source tree? |
| 00:11.01 | brlcad | well then, it looks like I maybe just implemented non-pushed matrix edit support for halfspaces |
| 00:12.27 | brlcad | depends on the platform, maybe half the time but I wouldn't say it's primarily in or out |
| 00:13.11 | brlcad | some platforms are almost exclusively out of dir (linux), some are mix in/out depending on what I'm doing (mac), and others still are almost always in dir (release) |
| 00:13.19 | starseeker | k - don't know why that would still matter but also something to keep in mind |
| 00:14.03 | brlcad | possible, but it was a pretty isolated failure with the double-class error |
| 00:14.35 | brlcad | i'll give it another test post-release |
| 00:16.02 | starseeker | brlcad: I think we found your new commuter car: http://blogs.wsj.com/tech-europe/2011/03/07/the-car-faster-than-a-speeding-bullet/ |
| 00:21.57 | starseeker | brlcad: yep: ERROR: NULL rt_half_internal pointer, file primitives/half/half.c, line 505 |
| 00:22.13 | starseeker | updates to confirm fix |
| 00:22.18 | brlcad | cool |
| 00:22.28 | brlcad | pretty much confirms, awesome |
| 00:22.49 | brlcad | hm, I don't see how/where oed actually applies the matrix |
| 00:31.10 | starseeker | brlcad: for which test? |
| 00:31.44 | starseeker | oh, confirmed that regression is fixed with the latest update |
| 00:32.58 | brlcad | that's bizzare though too |
| 00:33.10 | brlcad | the test does oed / oed_half.c/oed_half.s |
| 00:33.35 | brlcad | then merely presses accept |
| 00:33.47 | brlcad | that results in the halfspace import fail |
| 00:33.53 | brlcad | fishy |
| 00:33.57 | starseeker | O.o |
| 00:35.13 | brlcad | *sigh* .. back to gdb for understanding |
| 00:37.10 | starseeker | technically we don't have to pass those for release, they were never added to the "offical" regression testing... |
| 00:37.58 | brlcad | it's not a matter of passing the test, it's a failure related to code that changed |
| 00:38.23 | brlcad | the tests are simple enough, they're worth checking into to make sure something else wasn't horked |
| 00:38.57 | starseeker | nods - just noting that some of the "failures" may be nothing new... |
| 00:39.31 | brlcad | sure |
| 00:39.42 | brlcad | but they're the closest thing to a test at this point for all the librt code that was changed |
| 00:39.49 | starseeker | ah, point |
| 01:10.52 | starseeker | hmm... cmake build succeeds in running rtwizard and archer on 10.6, although I do see those kCGError messages |
| 01:11.03 | starseeker | will have to try autotools tomorrow |
| 01:15.27 | *** join/#brlcad Ralith (~ralith@S010600221561996a.vc.shawcable.net) |
| 02:48.01 | CIA-14 | BRL-CAD: 03brlcad * r43770 10/brlcad/trunk/NEWS: |
| 02:48.01 | CIA-14 | BRL-CAD: inadvertently reviewing and fixing bugs for release, fixed a never-implemented |
| 02:48.01 | CIA-14 | BRL-CAD: feature of halfspaces. if you put them into a combination and apply a matrix to |
| 02:48.01 | CIA-14 | BRL-CAD: the combination, librt would bomb with an invalid magic error. the half |
| 02:48.01 | CIA-14 | BRL-CAD: primitive was updated to handle that case which is used by mged to apply |
| 02:48.02 | CIA-14 | BRL-CAD: matrices to wireframes without writing to a disk record. |
| 02:49.00 | CIA-14 | BRL-CAD: 03brlcad * r43771 10/brlcad/trunk/TODO: technically unbusted though only because it's half-reverted. still investigating the cause, but todo is todone. |
| 03:16.09 | brlcad | gives bottie a final round of testing |
| 03:16.32 | starseeker | brlcad: the last report on bottie was "consistently crashing" |
| 03:17.47 | CIA-14 | BRL-CAD: 03brlcad * r43772 10/brlcad/trunk/src/librt/primitives/half/half.c: oops, wrong cast |
| 03:19.10 | brlcad | heh, k |
| 03:29.04 | brlcad | just worked with a simple sphere test case |
| 03:30.04 | brlcad | 50% speedup |
| 03:30.46 | brlcad | (of tie vs bot .. sph was around 200% faster) |
| 03:33.58 | starseeker | cool |
| 03:34.14 | starseeker | huh, this is interesting: http://www.cs.mtu.edu/~shene/COURSES/cs3621/NOTES/ |
| 03:34.16 | brlcad | nice 300% improvement on a model with just 9k tris |
| 03:34.25 | brlcad | 600k rtfm vs 200k rtfm |
| 03:35.15 | brlcad | I think I've seen that site |
| 03:37.16 | starseeker | needs to read through that |
| 03:37.18 | brlcad | and it scales nicely on a big image, 21 sec vs 7 sec |
| 03:37.23 | starseeker | awesome :-) |
| 03:37.26 | brlcad | the whole team should read through that |
| 03:37.41 | brlcad | about once a year ;) |
| 03:37.52 | starseeker | hehe |
| 03:38.24 | starseeker | will probably rediscover it around this time next year |
| 03:57.13 | brlcad | wow, actual 10x increase on a 0.5M poly bot (simple sphere) |
| 04:00.28 | brlcad | and on a 1.2M poly... |
| 04:02.41 | brlcad | 3 sec vs 62 sec |
| 04:02.58 | brlcad | hawt |
| 04:04.33 | brlcad | so that's about as good as it will probably get comparison-wise since it's a simple shape that's best for kdtree being compared with a bot not using pieces optimization (default) and is but a single simple shape |
| 04:09.00 | brlcad | nice, 5x increase on t62 |
| 04:09.13 | brlcad | (after removing the silly "I want normals" limitation) |
| 04:10.05 | brlcad | hm, maybe not on that one ... |
| 04:23.32 | brlcad | yeah, that wasn't right, looking to be about 40% faster (about 13s vs 21s) on a fully hierarchical bot t62 |
| 04:44.20 | *** join/#brlcad ibot (~ibot@rikers.org) |
| 04:44.20 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || #brlcad logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release prep for 7.18.2 under way (20110202) |
| 05:55.51 | CIA-14 | BRL-CAD: 03brlcad * r43773 10/brlcad/trunk/NEWS: looks like today is the day. reword the BoT-TIE metrics having directly observed (valid) performance from 40% to 20x faster. include details that the acceleration is presently disabled by default. |
| 05:58.37 | CIA-14 | BRL-CAD: 03brlcad * r43774 10/brlcad/trunk/TODO: peformance tested, results very promising |
| 06:00.02 | ``Erik | bottie breaks on 32b |
| 06:00.40 | ``Erik | took some work to get a 32b linux build (needs fixing), but it breaks there, too, not just fbsd |
| 06:01.55 | CIA-14 | BRL-CAD: 03brlcad * r43775 10/brlcad/trunk/src/librt/primitives/bot/bot.c: since it's disabled by default, is there really any point in requiring normals and orientation? tested on several unoriented and normal-free bots with success too. |
| 06:02.04 | ``Erik | (--disable-64bit-build causes failed build on a 64b linux build, -m32 is not being added correctly to src/other) |
| 06:02.22 | brlcad | yep, see README.Linux about that |
| 06:02.58 | brlcad | you basically have to force it kicking and screaming |
| 06:03.54 | ``Erik | heh, had to ./configure CFLAGS=-m32 CXXFLAGS=m32 and then force compile tk and another with /usr/lib/libfonconfig.so.1 instead of -lfontconfig |
| 06:04.25 | brlcad | probably because you need LDFLAGS-m32 too |
| 06:04.45 | ``Erik | no, there is no /usr/lib/libfontconfig.so ... mebbe an arl screwup |
| 06:05.02 | brlcad | probably, I manually fixed a couple bad X11 libs |
| 06:05.11 | brlcad | bad == missing symlink |
| 06:05.47 | brlcad | either way, results looking pretty good now |
| 06:05.53 | ``Erik | aaanyways, 32b tie crashes where I tried it, I've a lot of mods at work digging into it, but if you wanna dig in, knock yourself out... imma go unconcious |
| 06:06.25 | brlcad | i'm satisfied with it, I'm pressing for release tagging -- one last thing to check on |
| 06:06.51 | brlcad | it was working well enough to put numbers on the gains |
| 06:07.14 | brlcad | t62 is pretty darn close to real-world, it was around 40% |
| 06:07.15 | ``Erik | it takes a bit of work to trigger, so whatever... I'm not keen on release notes when it doesn't work 32b, but *shrug* I have more important stuff to worry about |
| 06:07.33 | brlcad | the notes say it's disabled by default and preliminary |
| 06:07.44 | brlcad | so there can be a future not when it's not off by default |
| 06:09.43 | CIA-14 | BRL-CAD: 03erikgreenwald * r43776 10/brlcad/trunk/TODO: note 32b issue for bottie |
| 06:10.21 | ``Erik | tomorrow; geomcore. |
| 06:12.16 | CIA-14 | BRL-CAD: 03brlcad * r43777 10/brlcad/trunk/TODO: huzzah! .. looks like the nmg failures were related to one of the other failures (probably nmg) because now everything is working again. |
| 07:00.49 | brlcad | tomorrow; tag. |
| 07:00.53 | brlcad | er, today |
| 07:46.36 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 08:28.59 | *** join/#brlcad epileg (~epileg@188.119.210.222) |
| 08:29.02 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 09:13.01 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 12:39.23 | *** join/#brlcad packrat (~packrator@99-67-225-40.lightspeed.livnmi.sbcglobal.net) |
| 12:51.07 | *** join/#brlcad Stattrav (~Stattrav@117.192.248.2) |
| 12:51.07 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 12:58.10 | *** join/#brlcad Stattrav (~Stattrav@122.167.250.138) |
| 12:58.10 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 12:59.37 | starseeker | reflects that a function to find all local maximums and minimums in a dsp dataset would probably be a good start at more intelligent spline surface generation... |
| 13:16.35 | *** join/#brlcad Stattrav (~Stattrav@122.167.250.138) |
| 13:16.35 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 13:22.07 | *** join/#brlcad Stattrav (~Stattrav@122.167.250.138) |
| 13:22.07 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 13:22.30 | *** join/#brlcad dli (~dli@dsl-173-248-211-229.acanac.net) |
| 13:31.37 | *** join/#brlcad Stattrav (~Stattrav@122.167.250.138) |
| 13:31.37 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 13:42.51 | *** join/#brlcad Stattrav (~Stattrav@117.192.248.2) |
| 13:42.51 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 13:49.34 | CIA-14 | BRL-CAD: 03starseeker * r43778 10/brlcad/branches/cmake/CMakeLists.txt: Add some documentation on what the distcheck routine does |
| 13:50.20 | CIA-14 | BRL-CAD: 03jordisayol * r43779 10/brlcad/trunk/ (misc/debian/rules sh/make_rpm.sh): change the number of simultaneous jobs on "make", during deb/rpm building process |
| 13:53.19 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 14:52.22 | ``Erik | hm, dns issues with brlcad.org ? |
| 15:04.18 | CIA-14 | BRL-CAD: 03brlcad * r43780 10/brlcad/trunk/misc/Makefile.am: debian/application-x-brlcad.png was replaced with two v4/v5 png files |
| 15:05.30 | brlcad | shouldn't be |
| 15:06.35 | ``Erik | I can't resolve from several different machines |
| 15:06.39 | ``Erik | including bz |
| 15:08.40 | brlcad | working for me |
| 15:08.55 | ``Erik | on bz? tried other machines to avoid cache artifacts? |
| 15:09.06 | epileg | ``Erik: not working here in Spain |
| 15:09.33 | brlcad | just reset named, try again |
| 15:12.08 | ``Erik | "no servers could be reached" |
| 15:12.52 | ``Erik | (from several different machines using different dns servers, other domains resolve) |
| 15:13.08 | brlcad | what query you using? |
| 15:13.19 | brlcad | does "nslookup google.com - bzflag.bz" work for you? |
| 15:13.25 | ``Erik | "nslookup brlcad.org" and "dig brlcad.org" |
| 15:13.41 | ``Erik | yes, that works |
| 15:14.43 | brlcad | oh my |
| 15:14.58 | epileg | here in spain, exactly same as ``Erik |
| 15:15.07 | ``Erik | on one machine, I got, um |
| 15:15.17 | ``Erik | ;; reply from unexpected source: 76.96.5.201#53, expected 68.87.71.230#53 |
| 15:15.20 | ``Erik | stuff like that |
| 15:15.46 | brlcad | could be related to zoneedit migration |
| 15:17.00 | ``Erik | *shrug* we'll see if it clears up, ah surpose |
| 15:19.16 | brlcad | yeah, NS1.ZONEEDIT.COM is not responding |
| 15:20.05 | brlcad | all three appear to be snookered |
| 15:20.08 | *** join/#brlcad ezzieyguywuf (~wolfie@cpe-071-070-255-232.nc.res.rr.com) |
| 15:20.12 | ``Erik | they don't have site redundant secondaries? O.o |
| 15:21.29 | ``Erik | ah, pay per secondary, lame :) |
| 15:22.15 | ``Erik | jabs some code for a while |
| 15:23.17 | brlcad | is paying for a secondary |
| 15:23.28 | brlcad | that's why there's three, one's even in the uk |
| 15:23.37 | brlcad | all three are unreachable |
| 15:23.56 | ``Erik | heh, 99.998, guess this is that .002 |
| 15:27.29 | ``Erik | irssi for noobs lessons *sigh* :D |
| 15:32.51 | brlcad | of 7 dns servers, looks like 5 of them are down |
| 15:35.08 | brlcad | er, 13 of 19 |
| 15:39.46 | starseeker | ``Erik: oh, don't worry - in another 10 years or so I may not be a noob anymore |
| 15:46.01 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 15:47.03 | ``Erik | so'z I gotta enjoy it while I can, right? :> |
| 15:47.21 | starseeker | heh |
| 15:49.10 | brlcad | note that brlcad.org is still reachable as bzflag.bz (at least for now) |
| 15:50.04 | brlcad | http://63.246.136.17/d/ will get you to the website |
| 15:50.19 | ``Erik | ayup, that's what I did earlier for my comic page, before calling out the dns issue (was poking around to see if the httpd was having issues, if'n ya look at the sudo log) |
| 15:51.13 | CIA-14 | BRL-CAD: 03erikgreenwald * r43781 10/geomcore/trunk/TODO: UUID is wired in and QT is gone |
| 15:58.19 | ``Erik | http://www.youtube.com/watch?v=xQqQ-Kcjowg "rental car olympics" |
| 16:12.21 | brlcad | heh |
| 16:21.27 | *** join/#brlcad Zaebos (~irc@pd95b7f5e.dip0.t-ipconnect.de) |
| 16:23.08 | *** join/#brlcad Nohla (~Nohla@64.76.19.227) |
| 16:33.58 | dli | brlcad, my first try for rt_revolve_xform(), http://pastebin.com/Tf3Qsg7J |
| 16:34.53 | dli | brlcad, still reading include/rtgeom.h |
| 16:57.27 | *** join/#brlcad Stattrav (~Stattrav@117.202.21.22) |
| 16:57.27 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 17:19.38 | starseeker | O.o |
| 17:25.11 | starseeker | What the... I'm seeing warning messages on OSX that are suppressed by adding -Werror |
| 17:34.32 | starseeker | no, that's not it... |
| 17:37.08 | starseeker | Ah - it's the presence or absence of STRICT_FLAGS in brlcad_config.h |
| 17:37.48 | starseeker | Ohhhhh. I see. |
| 17:38.22 | starseeker | bu.h linen 160 |
| 18:05.24 | CIA-14 | BRL-CAD: 03starseeker * r43782 10/brlcad/branches/cmake/ (6 files in 5 dirs): (log message trimmed) |
| 18:05.24 | CIA-14 | BRL-CAD: Rework how CFLAGS are assigned in CMake. Major changes are using build type |
| 18:05.24 | CIA-14 | BRL-CAD: specific flags to hold things instead of the catch-all, and changing the |
| 18:05.24 | CIA-14 | BRL-CAD: STRICT_FLAGS brlcad_config.h include to WARNING_FLAGS. The way CMake sets |
| 18:05.24 | CIA-14 | BRL-CAD: things up, turning on the warnings but not strict just means you're expecting to |
| 18:05.24 | CIA-14 | BRL-CAD: see all the same output without failing. Conditionalizing the bu.h |
| 18:05.24 | CIA-14 | BRL-CAD: undef/redefine logic on strict and not warning violated that principle, because |
| 18:18.32 | CIA-14 | BRL-CAD: 03starseeker * r43783 10/brlcad/branches/cmake/CMakeLists.txt: Setting linker flags now, not compiler flags |
| 18:20.45 | CIA-14 | BRL-CAD: 03starseeker * r43784 10/geomcore/trunk/tests/svntest/main.c: don't do a region subdirectory |
| 18:21.23 | CIA-14 | BRL-CAD: 03erikgreenwald * r43785 10/geomcore/trunk/src/utility/GSThread.cxx: destroy the mutex in the GSMutex destructor |
| 18:51.15 | starseeker | brlcad: I still can't reproduce the rtwizard/archer/bwish issue |
| 19:25.47 | *** join/#brlcad Nohla (~Nohla@64.76.19.227) |
| 19:45.24 | *** join/#brlcad Nohla (~Nohla@64.76.19.227) |
| 19:50.52 | CIA-14 | BRL-CAD: 03starseeker * r43786 10/brlcad/branches/cmake/ (10 files in 9 dirs): MFC r43785 |
| 20:01.18 | CIA-14 | BRL-CAD: 03starseeker * r43787 10/brlcad/branches/cmake/src/other/ (13 files in 6 dirs): Ah, right - src/other builds actually have to do their thing properly now. Get tkpng, tktable and tkhtml set up with some find_package calls. |
| 20:12.37 | CIA-14 | BRL-CAD: 03starseeker * r43788 10/geomcore/trunk/ (CMake/ CMakeLists.txt cmake/): Do like most other projects and put our CMake modules in a CMake directory |
| 20:21.28 | *** part/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 20:37.01 | CIA-14 | BRL-CAD: 03starseeker * r43789 10/geomcore/trunk/ (3 files in 2 dirs): Make realpath happy on OSX |
| 20:49.58 | CIA-14 | BRL-CAD: 03jordisayol * r43790 10/brlcad/trunk/ (3 files in 2 dirs): add an "update gtk icon cache" and some minor improvements on "install" and "remove" scripts of deb/rpm packages |
| 20:59.53 | *** join/#brlcad dli_ (~dli@dsl-69-171-148-245.acanac.net) |
| 21:01.21 | CIA-14 | BRL-CAD: 03bob1961 * r43791 10/brlcad/trunk/src/tclscripts/lib/TkTable.tcl: Added a TkTable::see method. |
| 21:06.09 | CIA-14 | BRL-CAD: 03erikgreenwald * r43792 10/geomcore/trunk/src/libNet/NetMsgRouter.cxx: update getListOfHandlers to work correctly with STL maps |
| 21:20.05 | CIA-14 | BRL-CAD: 03brlcad * r43793 10/brlcad/trunk/src/libged/gqa.c: |
| 21:20.06 | CIA-14 | BRL-CAD: add some error recovery to gqa so that we don't bomb out during |
| 21:20.06 | CIA-14 | BRL-CAD: bu_malloc/bu_calloc when passed a zero size allocation. probably means |
| 21:20.06 | CIA-14 | BRL-CAD: something earlier went awry but check here regardless so we can be more graceful |
| 21:20.06 | CIA-14 | BRL-CAD: about halting. |
| 21:20.29 | starseeker | ``Erik: that's got it, thanks! |
| 22:34.36 | CIA-14 | BRL-CAD: 03starseeker * r43794 10/geomcore/trunk/tests/svntest/main.c: Put each geometry object into its own directory. |
| 22:39.03 | starseeker | brlcad: about 50 seconds when I put each object in its own directory (regions and assemblies) |
| 23:10.30 | CIA-14 | BRL-CAD: 03starseeker * r43795 10/brlcad/branches/cmake/include/bu.h: Go ahead and keep the old comment |
| 23:13.47 | CIA-14 | BRL-CAD: 03starseeker * r43796 10/brlcad/trunk/ (configure.ac include/bu.h): STRICT_FLAGS -> WARNING_FLAGS - just a rename as far as autotools is concerned, no behavior change |
| 23:45.26 | CIA-14 | BRL-CAD: 03starseeker * r43797 10/geomcore/trunk/src/GS/ (9 files): First step of renaming geoclient and geoserv to geomclient and geomserv |
| 23:47.36 | CIA-14 | BRL-CAD: 03starseeker * r43798 10/geomcore/trunk/src/GS/ (geomclient.cxx geomserv.cxx): geoclient and geoserv renamed to use geom prefix - sounds less like geospatial related software |