| 00:33.55 | CIA-48 | BRL-CAD: 03brlcad * r41293 10/brlcad/trunk/configure.ac: few more warnings to consider |
| 01:58.55 | *** join/#brlcad R0b0t1 (~Enigma@unaffiliated/r0b0t1) |
| 07:43.00 | *** join/#brlcad stevegt_ (~stevegt@c-69-181-134-76.hsd1.ca.comcast.net) |
| 08:37.11 | *** join/#brlcad mafm (~mafm@83.37.154.171) |
| 09:37.05 | *** join/#brlcad WhiteCalf (MK@whitecalf.net) |
| 09:42.23 | *** join/#brlcad Ralith_ (~ralith@S010600221561996a.vc.shawcable.net) |
| 11:01.02 | CIA-48 | BRL-CAD: 03indianlarry * r41294 10/brlcad/trunk/src/librt/opennurbs_ext.cpp: fixed delete statement for allocated array (delete knots -> delete [] knots) |
| 11:02.38 | CIA-48 | BRL-CAD: 03indianlarry * r41295 10/brlcad/trunk/src/librt/bundle.c: removed unused static debug variable |
| 11:11.52 | CIA-48 | BRL-CAD: 03indianlarry * r41296 10/brlcad/trunk/src/librt/primitives/brep/brep.cpp: (log message trimmed) |
| 11:11.52 | CIA-48 | BRL-CAD: Fixed under sized arrays used in reporting intersects in four corner solver. |
| 11:11.52 | CIA-48 | BRL-CAD: This under size was causing stack corruption issues when a subdivision surface |
| 11:11.52 | CIA-48 | BRL-CAD: was hit more than twice. Added definition MAX_BREP_SUBDIVISION_INTERSECTS and |
| 11:11.52 | CIA-48 | BRL-CAD: currently set to 5, four corners and center are currently being tested for |
| 12:01.27 | d-lo | yawns |
| 12:08.45 | brlcad | wow, so that's why nmg conversion of havoc is taking so long.. there's like an O(n^5) in there |
| 12:13.15 | brlcad | n^6 or n^8 even |
| 12:13.27 | brlcad | incredible |
| 12:14.05 | d-lo | n7 guys in Mass Effect were pretty elite.... but thats not what you're talking about ;) |
| 12:16.59 | d-lo | sorry for the total tangent. :) |
| 12:17.16 | d-lo | how easy a fix is that to get closer to ^2 or linear? |
| 12:26.04 | CIA-48 | BRL-CAD: 03brlcad * r41297 10/brlcad/trunk/ (NEWS src/librt/primitives/nmg/nmg_fuse.c): |
| 12:26.04 | CIA-48 | BRL-CAD: this fixes an infinite loop that was getting hit when fusing vertices due to |
| 12:26.04 | CIA-48 | BRL-CAD: some faulty iteration logic. instead of continuing if they are not equal, which |
| 12:26.04 | CIA-48 | BRL-CAD: was causing it to stay in the same processing state, check if they are equal so |
| 12:26.04 | CIA-48 | BRL-CAD: we can fuse and break. it was stuck endlessly fusing. affects most NMG and BOT |
| 12:26.04 | brlcad | trivial |
| 12:26.21 | brlcad | it was just carelessly being too careful |
| 12:26.39 | d-lo | nice :) Easy fixes are the best. |
| 12:27.03 | d-lo | do you have benchmarks as to how much improvement your fix gained? |
| 12:27.24 | brlcad | not yet |
| 12:27.55 | brlcad | part of that nmg conversion script I whipped up, has exposed a variety of problems |
| 12:28.07 | d-lo | well awesome then. |
| 12:28.19 | d-lo | assuming those problems get fixed of course :) |
| 12:28.19 | brlcad | it's been busy processing every object in db/*.g |
| 12:28.34 | brlcad | every now and then it hits an object and takes .. a very very long time |
| 12:29.26 | brlcad | I've been attaching the debugger to make sure it's not an infinite loop (which one WAS) .. but now ran into a case where the time complexity wasn't just geometric |
| 12:30.00 | brlcad | relatively simple piece of geometry, debugger showed it wasn't stuck, found crazyness |
| 12:30.18 | d-lo | now is this nmg stuff ours, but old? |
| 12:30.33 | brlcad | there's an O(n^2) NMG_CK_HITMISS_LISTS() macro that validates that the entries in a hitmiss list point to valid memory |
| 12:30.44 | brlcad | that check is fine, but it was *everywhere* |
| 12:31.15 | brlcad | so it was iterating over the hitmiss list as it walked down the same callstack even |
| 12:31.23 | d-lo | hrm, alsmost sounds like that portion of code never had the 'debugness' of it taken out. |
| 12:32.00 | brlcad | f() -> g() -> h() -> i() basically calling it every step down at the beginning and end of every function |
| 12:32.12 | brlcad | so doing more than 10x the amount of work it needs to, only to test memory |
| 12:32.46 | d-lo | so did you just gut that stuff or put some dort of debug flag around it? |
| 12:32.59 | brlcad | unrolled them properly up the stack |
| 12:33.18 | brlcad | so only f() does the check before and after calling g() |
| 12:33.23 | d-lo | kewl :) |
| 12:34.45 | d-lo | is that call stack the same every time? Cause I wonder if only having the memcheck in f() is completely safe.... |
| 12:34.51 | d-lo | *just thinking abstractly* |
| 12:34.57 | d-lo | as I havent seen the code. |
| 12:41.24 | *** join/#brlcad juanman (~quassel@unaffiliated/juanman) |
| 12:41.36 | ``Erik | hmmm |
| 12:54.58 | brlcad | they're all safe checks |
| 13:23.29 | *** join/#brlcad R0b0t1 (~Enigma@unaffiliated/r0b0t1) |
| 13:38.17 | brlcad | deprecates 40 commands |
| 13:53.25 | CIA-48 | BRL-CAD: 03brlcad * r41298 10/brlcad/trunk/doc/deprecation.txt: (log message trimmed) |
| 13:53.26 | CIA-48 | BRL-CAD: massive deprecation of various mged commands that are being consolidated into a |
| 13:53.26 | CIA-48 | BRL-CAD: more consistent interface with fewer commands. NO FUNCTIONALITY IS BEING |
| 13:53.26 | CIA-48 | BRL-CAD: REMOVED, they are being merged in with other existing commands as options or |
| 13:53.26 | CIA-48 | BRL-CAD: subcommands so common functionality is grouped together. this should eliminate |
| 14:06.42 | CIA-48 | BRL-CAD: 03brlcad * r41299 10/brlcad/trunk/doc/deprecation.txt: |
| 14:06.42 | CIA-48 | BRL-CAD: add a few more easy ones to include the list for deprecation. remove need to |
| 14:06.42 | CIA-48 | BRL-CAD: distinguish between prims and combs (no 'o' commands or options). deprecaters |
| 14:06.42 | CIA-48 | BRL-CAD: include: aproposlib aproposdevel qvrot qorot sed oed ted red vdraw vnirt |
| 14:06.42 | CIA-48 | BRL-CAD: vquery_ray |
| 14:06.50 | brlcad | inital list truncated was: ?lib ?devel dbconcat dbupgrade dbfind dbupgrade debugbu debugdir debuglib debugmem debugnmg comb_std comb_color combmem killall killtree killrefs rotobj arot mrot orot vrot oscale pscale otranslate ptranslate nmg_collapse nmg_fix_normals nmg_simplify bot_condense bot_decimate bot_face_fuse bot_vertex_fuse bot_face_sort bot_smooth erase_all query_ray xpush |
| 14:07.10 | brlcad | so that'll be about 50 commands that are easy to refactor |
| 14:07.18 | brlcad | shoudl be a good start |
| 14:13.40 | CIA-48 | BRL-CAD: 03brlcad * r41300 10/brlcad/trunk/NEWS: (log message trimmed) |
| 14:13.40 | CIA-48 | BRL-CAD: keith fixed a bug in the NURBS raytracing code where it was crashing on a |
| 14:13.40 | CIA-48 | BRL-CAD: particular surface evaluation. the ray was grazing a surface and the code was |
| 14:13.40 | CIA-48 | BRL-CAD: assuming only two hits were possible (in/out) when the surface was flat. this |
| 14:13.40 | CIA-48 | BRL-CAD: particular crash case involved three hits causing the fixed-size array bounds to |
| 14:16.20 | CIA-48 | BRL-CAD: 03starseeker * r41301 10/brlcad/branches/STABLE/ (62 files in 14 dirs): Sync STABLE branch to trunk r41297. |
| 14:35.06 | d-lo | Great stuff. And SFW: http://i.imgur.com/uaOrr.jpg |
| 15:53.51 | CIA-48 | BRL-CAD: 03starseeker * r41302 10/brlcad/trunk/doc/README.MacOSX: Toss in a note in case I need to remember how to query what version of OSX I'm using from the command line. |
| 15:57.43 | CIA-48 | BRL-CAD: 03starseeker * r41303 10/brlcad/trunk/doc/README.Linux: Mention the version file for Redhat and Fedora while we're at it. |
| 16:02.32 | CIA-48 | BRL-CAD: 03bob1961 * r41304 10/brlcad/trunk/src/tclscripts/mged/mgedrc.tcl: Fixed a typo in the dump_mged_state proc related to the status_bar default. |
| 16:11.03 | CIA-48 | BRL-CAD: 03bob1961 * r41305 10/brlcad/trunk/src/tclscripts/mged/bindings.tcl: |
| 16:11.03 | CIA-48 | BRL-CAD: Modified the default_mouse_bindings proc by replacing the previous hack, related |
| 16:11.03 | CIA-48 | BRL-CAD: to bad mouse behavior on Mac OS 10.5, with a similar hack that let's the user |
| 16:11.03 | CIA-48 | BRL-CAD: determine whether or not to use the provided hack by setting a no_focus_hack |
| 16:11.03 | CIA-48 | BRL-CAD: variable in the .mgedrc file. |
| 16:17.14 | *** join/#brlcad Zaebos (~irc@pd95b7f5e.dip0.t-ipconnect.de) |
| 17:11.30 | *** join/#brlcad 14WAABF48 (~stevegt@2001:470:1f05:601:212:f0ff:fe44:bea2) |
| 19:35.39 | d-lo | Well now! http://www.cbsnews.com/stories/2010/11/09/national/main7036716.shtml |
| 19:52.56 | *** join/#brlcad Ralith (~ralith@d142-058-092-131.wireless.sfu.ca) |
| 19:53.52 | brlcad | yeah, interesting |
| 19:54.37 | d-lo | Somewhere, on some sub, there is a junior Fire Control Tech who is no longer a Fire Control Tech |
| 19:54.57 | brlcad | heh |
| 19:57.39 | CIA-48 | BRL-CAD: 03brlcad * r41306 10/brlcad/trunk/ (NEWS src/librt/primitives/nmg/nmg_rt_isect.c): (log message trimmed) |
| 19:57.40 | CIA-48 | BRL-CAD: undo code that is carelessly being too careful. NMG_CK_HITMISS_LISTS() is a |
| 19:57.40 | CIA-48 | BRL-CAD: linear walk over the hit and miss lists that was being called all over the place |
| 19:57.40 | CIA-48 | BRL-CAD: up and down the call stack, often very redundantly. this is causing certain |
| 19:57.40 | CIA-48 | BRL-CAD: objects that need to fire rays to evaluate the nmg surface to run absurdly |
| 20:42.38 | starseeker | d-lo: hehe - "Wait, don't set that cup there- oh, crap" |
| 20:45.33 | starseeker | aaaand it looks like our RPM build is busted |
| 20:46.13 | CIA-48 | BRL-CAD: 03bob1961 * r41307 10/brlcad/trunk/src/libged/rt.c: Modified ged_build_tops to use the full path name when building up the command line vector. |
| 20:54.16 | CIA-48 | BRL-CAD: 03starseeker * r41308 10/brlcad/trunk/misc/brlcad.spec.in: Evidently Copyright is supposed to be License these days. |
| 21:45.32 | CIA-48 | BRL-CAD: 03bob1961 * r41309 10/brlcad/trunk/src/libged/tables.c: Need to dereference numreg and numsol before incrementing. |
| 21:51.05 | CIA-48 | BRL-CAD: 03starseeker * r41310 10/brlcad/trunk/NEWS: Bob fixed bug in idents, solids and regions command where counting was not functioning due to an improperly referenced counter. |
| 22:05.18 | *** join/#brlcad velociostrich (~nsd@c-68-37-119-2.hsd1.nj.comcast.net) |
| 23:18.24 | *** join/#brlcad stevegt_ (~stevegt@216.1.176.121) |
| 23:47.46 | *** join/#brlcad velociostrich (~nsd@c-68-37-119-2.hsd1.nj.comcast.net) |
