| 00:03.31 | *** join/#brlcad cad36 (n=3a975ca7@bz.bzflag.bz) |
| 00:06.36 | *** join/#brlcad cad36 (n=3a975ca7@bz.bzflag.bz) |
| 00:37.47 | *** join/#brlcad poolio_ (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 01:12.09 | *** join/#brlcad poolio_ (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 01:24.14 | *** join/#brlcad poolio (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 01:37.51 | *** join/#brlcad poolio_ (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 01:52.47 | *** join/#brlcad poolio (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 04:09.30 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/TODO: |
| 04:09.30 | CIA-4 | BRL-CAD: add a deprecation list for keeping track of when things can be marked obsolete |
| 04:09.30 | CIA-4 | BRL-CAD: and subsequently removed. the general guidelines are to allow removal during a |
| 04:09.30 | CIA-4 | BRL-CAD: major update, after at least three minor updates, or after six months .. |
| 04:09.30 | CIA-4 | BRL-CAD: whichever comes first. |
| 04:10.28 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/librt/bomb.c: rt_bomb() has technically been deprecated since before 7.0, but even our sources only loosely reflected this via inclusion in compat4.h so begin deprecation now, adding a deprecation message. |
| 04:14.38 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/ (110 files in 16 dirs): rt_bomb() has been deprecated since before 7.0, but even our sources weren't properly updated to reflect this change. update all of our sources to use bu_bomb() now. |
| 05:05.10 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/autogen.sh: include a contact e-mail |
| 07:32.34 | *** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch) |
| 09:00.21 | *** join/#brlcad elite01 (n=elite01@dslc-082-082-083-217.pools.arcor-ip.net) |
| 09:15.13 | *** join/#brlcad Elperion (n=Elperion@p54877b0d.dip.t-dialin.net) |
| 09:28.25 | *** join/#brlcad Elperion (n=Bary@p54877b0d.dip.t-dialin.net) |
| 09:43.03 | *** join/#brlcad Elperion (n=Bary@p54877b0d.dip.t-dialin.net) |
| 13:06.21 | CIA-4 | BRL-CAD: 03bob1961 * 10brlcad/src/tclscripts/archer/Archer.tcl: Replace "get" with "get_type" for all cases where it was being used to determine the existence of an object. |
| 13:35.15 | *** join/#brlcad Elperion (n=Bary@p54877b0d.dip.t-dialin.net) |
| 14:29.06 | brlcad | I'd made a mental note to check out other polynomial root solvers to see if we can get some free performance boosting to the more complex primitives that use it for solving their ray-intersection |
| 14:29.31 | brlcad | i finally got around to testing some today, and have tested four so far that I've run across .. |
| 14:30.07 | brlcad | not one has come even close to the performance .. closest has been about 300% slower |
| 14:31.28 | archivist | everone out there re-inventing slower wheels |
| 14:32.09 | brlcad | i actually didn't think ours was "that fast" .. |
| 14:33.39 | archivist | a nice shock then |
| 14:33.42 | brlcad | I knew it was good, but nothing ground breaking .. i'm just not even finding anything in the ballpark |
| 14:34.01 | brlcad | well, it could just be that I've just run across four lemons in a row too |
| 14:34.14 | brlcad | but so far yeah.. not too shabby for code that old.. |
| 14:35.48 | ``Erik | y'know, I'm under the impression that older code tends to be MUCH better performance wise |
| 14:35.54 | ``Erik | both asymoptotic and constant |
| 14:35.55 | ``Erik | :/ |
| 14:36.45 | brlcad | jeff hanes wrote the existing solver back in '85 |
| 14:36.46 | ``Erik | machines were slower, they couldn't afford the waste... take a look at regex nfa's... ruby, perl, etc are all abysmal compared to early 70's thompson nfa's (grep, sed), and even tcl blows the doors off of perl for 'difficult' regex's |
| 14:37.42 | ``Erik | got a paper on my desk... lemme find the url... |
| 14:37.46 | brlcad | i was sorta expecting to find a simd'd root solver somewhere that might have a good hand at competing, but so far no luck |
| 14:38.10 | ``Erik | http://swtch.com/~rsc/regexp/regexp1.html |
| 14:38.41 | ``Erik | I THINK I got the paper from http://lambda-the-ultimate.org/node/2064 |
| 14:38.49 | archivist | heh I look at some modern database stuff and the mid 1980s stuff had some very good ideas that now seem forgotten |
| 14:40.04 | brlcad | ``Erik: oh, i'd fully believe that -- I ran into regex issues when I wrote this extensive language filter for bzflag -- expressions that were hundreds of chars long, complex patterns |
| 14:40.16 | archivist | early 60's bah /me has a 1951 programming book on my desk |
| 14:40.22 | ``Erik | heh |
| 14:40.45 | archivist | actually a M.V. Wilkes book |
| 14:40.52 | brlcad | i ended up using the (old) version we have in cad on some systems because it would crash various newer system regex libraries (debian was one) |
| 14:40.53 | ``Erik | brlcad: check out the url I pasted, it might be worth your time :) |
| 14:42.01 | ``Erik | a 30 character match taking 60 seconds using perl regex and 20 seconds using thompson nfa in awk/grep |
| 14:42.03 | ``Erik | er |
| 14:42.06 | ``Erik | 20 microseconds |
| 14:42.32 | ``Erik | .00002 seconds |
| 14:43.04 | brlcad | yeah, i'm not really too surprised by that |
| 14:43.24 | brlcad | though I was only considering implementations already in the lower graph |
| 14:44.18 | brlcad | my regex's were hundreds of chars.. and I needed it to test thousands within less than 10ms |
| 14:44.50 | brlcad | i had all the timings worked out, it was rather expensive |
| 14:45.23 | ``Erik | heh, (of course, this is the pathological case... which is the word I was searching for yesterday...) |
| 14:47.28 | brlcad | for the root solving, it could also be the case that all/most were based off some reference implementation that everyone just keeps copying, maybe from graphics gems or something |
| 14:49.30 | ``Erik | brlcad: working from home today? |
| 14:50.12 | brlcad | just not working |
| 14:50.33 | ``Erik | ah :) guess I'll stay in for luncht hen :D |
| 14:50.51 | brlcad | give me some time to poke on some more interesting code stuffs (like the root finding) |
| 14:51.16 | ``Erik | opposed to bzflag stuff? :D |
| 14:51.45 | brlcad | that too |
| 14:53.49 | brlcad | heh, most are |
| 14:54.48 | ``Erik | mebbe after my sexp/xml compilers, heh |
| 14:57.00 | brlcad | bah.. paper not included in my ieee membership |
| 14:57.24 | ``Erik | hum, think I could get it through the portal? |
| 14:57.46 | brlcad | http://ieeexplore.ieee.org/iel5/10248/32586/01524367.pdf?arnumber=1524367 |
| 14:59.03 | ``Erik | /tmp on you know where |
| 15:09.16 | brlcad | shweet |
| 15:10.16 | brlcad | gah, they're solving arbitrary degree polynomials |
| 15:10.28 | ``Erik | (but only since it's explicitely for a BRL-CAD task and all official like that, 'n shtuff, you coulda ran firefox on a work box to get it... :D ) |
| 15:10.30 | brlcad | don't care much past 4th |
| 15:10.47 | ``Erik | hum, at least not for the torus |
| 15:10.56 | brlcad | torus is 4th |
| 15:11.14 | ``Erik | if the meatball thing were converted to a root solve, it'd be nth degree |
| 15:11.53 | ``Erik | nurbles might leverage >4, I'd imagine... but I don't know jack about those |
| 15:12.16 | brlcad | the existing will go higher than 4, it's just a compile-time limit |
| 15:12.51 | ``Erik | one of my test cases for the brute force approach was 10,000 control points, which I THINK is a 10k order polynomial |
| 15:13.16 | brlcad | meatballs are good to keep in mind .. though if you have them working direct without root solver, that's probably a better solution anyways |
| 15:13.36 | ``Erik | I have an approximation working, it's imperfect |
| 15:13.56 | brlcad | depends how imperfect :) |
| 15:14.05 | ``Erik | um, at the moment, it walks the ray in 1/20 bounding sphere radius steps looking for a sign change, effectively |
| 15:14.07 | brlcad | i mean they are metaballs after all anyways too |
| 15:14.21 | ``Erik | then does a binary partition to find the 'close enough' point |
| 15:14.29 | ``Erik | within like 1/1000 radius |
| 15:14.31 | brlcad | eek.. a semi-newton iteration |
| 15:14.36 | ``Erik | ayup |
| 15:15.01 | ``Erik | so if the volume were between initial steps, it'd be ignored |
| 15:15.08 | ``Erik | which gets edge artifacts :/ |
| 15:15.23 | brlcad | how does it deal with the control points? |
| 15:15.30 | brlcad | iterates over each? |
| 15:16.25 | ``Erik | um, $v_{V}=\Sigma_{i}\frac{f_{i}}{|V-p_{i}|}$ |
| 15:17.05 | ``Erik | src/librt/g_metaball.c:203 |
| 15:17.06 | brlcad | yeah.. that's not useful to me :) |
| 15:17.20 | ``Erik | and :220 |
| 15:17.39 | brlcad | ah, so it does iterate over the points |
| 15:17.42 | ``Erik | (shut up, it's ugly, but it works) |
| 15:17.42 | ``Erik | yes |
| 15:18.02 | brlcad | looks like it always does shoot-through too |
| 15:18.08 | ``Erik | has to |
| 15:18.18 | ``Erik | you don't tknow how many chunks of volume it'll go through |
| 15:18.28 | ``Erik | well, heh, n/m |
| 15:18.55 | ``Erik | let me re-read, been a while since I've looked |
| 15:19.14 | ``Erik | no, I think it stops on first hi, always |
| 15:19.52 | ``Erik | no, wait, I do have one-hit in there, :259 |
| 15:20.53 | brlcad | ah, there it is.. first line in shot() |
| 15:21.55 | ``Erik | (bear in mind, the "tactical" purpose for the work was to hook directly to the point evaluator, the shot is just visual candy) :/ |
| 15:22.11 | brlcad | hm, would actually be interesting to see how the root solver fairs |
| 15:22.29 | brlcad | i'd suspect it'd have trouble converging on arbitrary large-point metas |
| 15:22.41 | brlcad | mebbie, dunno |
| 15:22.44 | clock_ | brlcad: the rendering seems to be extremely slow if screws cut in half are present in the scene, even when they take up just a tiny amount of pixels |
| 15:23.16 | ``Erik | one of the future enhancements was to go beyond point and handle line and tri-mesh, perhaps totally arbitrary geometry... :/ |
| 15:23.31 | ``Erik | which'd greatly complicate the root solve method |
| 15:23.35 | brlcad | clock_: it's shouldn't depend on the amount of pixels -- it would be how long it spends on those pixels |
| 15:24.05 | brlcad | having them in the scene is going to slow down neighboring pixels some too due to space partitioning |
| 15:24.11 | clock_ | brlcad: I made the screw a group of turns, each turn is a group of slanted cylinders. So it has a hierarchical structure |
| 15:25.02 | ``Erik | in the geometry tree, it may b e clean... but the raytracing uses a KD-tree, too |
| 15:25.06 | brlcad | your csg tree might be unoptimized too, which could be slowing things way down |
| 15:26.21 | brlcad | ``Erik: do you know anything about /var/tmp disappearing on .bz? |
| 15:26.26 | ``Erik | no? |
| 15:26.38 | brlcad | it's been getting deleted automatically for a couple days now |
| 15:26.39 | ``Erik | anything in the sudo log? |
| 15:26.41 | ``Erik | ugh |
| 15:26.42 | brlcad | nothing |
| 15:26.51 | ``Erik | grep -ri /var/tmp /etc |
| 15:26.52 | ``Erik | ? |
| 15:27.14 | ``Erik | also; sudo crontab -l |
| 15:27.15 | brlcad | yup, nothing of interest |
| 15:27.22 | brlcad | also checked |
| 15:27.28 | ``Erik | hum, /var/tmp should only be deletable by r00tus r00tus |
| 15:27.44 | ``Erik | none of my boxen of that os lose their /var/tmp :/ |
| 15:27.49 | brlcad | exactly -- i was (hoping) you'd set something up to see what all would break |
| 15:28.15 | ``Erik | no, if I'm commiting crazy break testing, I do it on a machine I can walk to and physically touch |
| 15:28.35 | ``Erik | if you think I'm willing to drive to florida to log in on console and fix something, you need to pass that pipe O.o :D |
| 15:28.42 | brlcad | i've got a watch thread going now to see if/when it disappears |
| 15:28.51 | brlcad | it's been happening for like three days now |
| 15:29.21 | ``Erik | the only thing I've done on that machine in the last week or two is log in, run irssi for a minute, kill irssi and log out |
| 15:30.37 | brlcad | i made a symlink this time, no particular reason |
| 15:31.06 | brlcad | only thing suspicious is rootkit attempts via http |
| 15:31.11 | brlcad | to /var/tmp |
| 15:31.27 | ``Erik | that'd be bad |
| 15:31.50 | brlcad | but they seemed rather benign .. they download a linux binary and try to run it |
| 15:32.04 | ``Erik | um, linuxulator is off, right? |
| 15:32.08 | brlcad | it doesn't get that far, of course, though even if it did, it'd fail (tested it ;) |
| 15:32.14 | brlcad | right |
| 15:32.44 | ``Erik | ummmmm, got 'fam' on the machine? |
| 15:32.48 | ``Erik | or tripwire? |
| 15:33.05 | brlcad | so i'm thinking maybe some php intrusion is getting through.. but the worst it can do is delete it's tmpdir |
| 15:33.49 | ``Erik | but only root should be able to rm that dir, and everything php/httpd should touch should be www:www |
| 15:33.50 | brlcad | not at the moment, though pieces of what those do are running |
| 15:34.41 | ``Erik | might be worth putting a tail -f on some /var/log stuff that dumps to another machine |
| 15:34.42 | brlcad | i'm just reaching.. it's the only hint that even generally involved /var/tmp |
| 15:35.54 | brlcad | i'm going to wait for the while test to fail.. so at least i'll have a timeframe that I can hunt for in all the logs |
| 15:36.13 | ``Erik | aight *shrug* |
| 15:36.36 | ``Erik | if it's integrity related, that's serious |
| 15:38.37 | brlcad | i'm not finding anything that indicates one way or the other so far, just guessing |
| 15:39.00 | brlcad | just like asking you was a guess that you might be up to something, however unlikely |
| 15:39.14 | ``Erik | nope, not my doing |
| 15:39.38 | brlcad | noticed /usr/tmp being added/removed for portupgrade through the searching, figued "maybe" |
| 15:40.35 | ``Erik | /usr/tmp was added cuz it's "normal" to have it, and removed cuz you complained... if portupgrade is doing it, *shrug* I d'no... but I've been using portmanager and those all leave sudo footprints |
| 15:41.04 | *** join/#brlcad Elperion (n=Bary@p54877b0d.dip.t-dialin.net) |
| 15:48.20 | ``Erik | hrmph |
| 15:53.04 | brlcad | i don't see why i'd even care |
| 15:53.33 | ``Erik | hrm, you didn't want that dir there "to see what would break" iirc |
| 15:53.56 | ``Erik | wanna grab lunch somewhere? I'm trying to decide whether to bust open a can of soup or not :/ |
| 15:54.02 | brlcad | that was more "it'd be interesting to see what breaks" .. though nothing did |
| 15:54.11 | brlcad | didn't really care either way to begin with |
| 15:54.39 | ``Erik | <-- tries to limit his involvement with that machine |
| 15:56.02 | brlcad | *shrug* |
| 15:56.51 | brlcad | ah, i already started munching on a sandwhich about an hour ago when someone mentioned food |
| 15:57.19 | ``Erik | aight, canned stuff it is :) I'm hitting the grocery store tonight, so'z I can replenish my supplies, don't gotta worry about being down to the last can |
| 15:58.35 | ``Erik | heh, I'm sure mm woulda loaned you a cup o' newdlez |
| 15:58.54 | ``Erik | <-- prefers keeping cans of progresso, is getting old :( |
| 15:58.58 | brlcad | not my fav |
| 15:59.22 | brlcad | i was on a progressor fix a few years ago .. but that died down after a few months |
| 15:59.56 | brlcad | ramen's never died down though.. i turn those babies into gourmet |
| 16:04.44 | ``Erik | heh |
| 16:05.45 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/configure.ac: disable unnecessary libtool test for fortran |
| 16:06.37 | ``Erik | (do we have any fortran?) |
| 16:07.14 | ``Erik | f2c would be handy for "those other guys" |
| 16:16.23 | brlcad | we provide one example file on how to shoot a ray in fortran |
| 16:17.30 | brlcad | so .. yes we do, but we still don't need a fortran compiler |
| 16:17.30 | ``Erik | hum, I see two .f files |
| 16:17.30 | brlcad | oh yeah, one in patch too |
| 16:17.36 | brlcad | also not used |
| 16:18.02 | ``Erik | if the conv/patch one just a remnant? can it be rm'd? |
| 16:18.22 | brlcad | yeah |
| 16:18.39 | brlcad | just a helper tool to much with patch input files |
| 16:18.44 | brlcad | s/much/muck/ |
| 16:32.55 | brlcad | wow, the GNU scientific library's solver is horribly slow |
| 16:34.51 | brlcad | gsl is 220 seconds vs libbn's 9 seconds |
| 16:35.10 | ``Erik | heh |
| 16:35.49 | brlcad | i had higher hopes for them .. |
| 16:36.13 | brlcad | theirs may be, however, significantly more robust |
| 16:36.49 | brlcad | that could certainly be affecting numbers.. but that is still a huge gap |
| 16:37.15 | brlcad | best so far has been about 30 seconds |
| 16:37.30 | brlcad | (solving 50M random 4th degree polynomials) |
| 16:46.58 | ``Erik | sweet, my last car payments looks like it'll be around july 5th |
| 16:47.08 | ``Erik | s/s / / |
| 16:55.02 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/TODO: change root solver to have coefficients in the right order |
| 18:05.59 | CIA-4 | BRL-CAD: 03jlowenz * 10brlcad/src/conv/iges/PullbackCurve.cpp: implement b-spline interpolation: generating knots and parameters. next up: generating the control points. |
| 18:37.38 | CIA-4 | BRL-CAD: 03jlowenz * 10brlcad/src/conv/iges/PullbackCurve.cpp: convert use of vector to TNT arrays. implement control point generation using JAMA's LU decomposition. |
| 18:48.49 | CIA-4 | BRL-CAD: 03jlowenz * 10brlcad/src/conv/iges/PullbackCurve.cpp: implement nurbs curve generation. an attempt at simplification (test for arc?) should be done at some point. |
| 19:16.06 | *** join/#brlcad clock_ (i=clock@77-56-106-32.dclient.hispeed.ch) |
| 19:52.47 | brlcad | well, finally found one that is slightly faster, but it fails to produce a solution |
| 19:54.05 | ``Erik | "My diet of McDonalds and pizza keeps me regular. I take one huge, painful, bloody dump every four days; regular as clockwork." |
| 19:55.04 | brlcad | heh |
| 19:55.27 | ``Erik | (yes, I'm reading poll comments on slashdot. :( ) |
| 19:55.51 | ``Erik | "When I wake up in the morning, I just can't get started until I've had that first, piping hot pot of coffee. Oh, I've tried other enemas..." |
| 20:02.35 | archivist | hehe |
| 20:06.07 | *** join/#brlcad poolio (n=poolio@c-69-251-3-107.hsd1.md.comcast.net) |
| 20:12.01 | brlcad | hah, so it was just slightly faster, because it was calling an _empty_ base class.. that was all overhead |
| 20:19.27 | ``Erik | neat, saw a couple broad-headed skinks infront of the building |
| 20:21.36 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/libbn/poly.c: commentification |
| 20:36.24 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/TODO: refactor complex.h and plane.h |
| 20:58.55 | *** join/#brlcad louipc (n=louipc@bas8-toronto63-1096734947.dsl.bell.ca) |
| 21:49.06 | *** join/#brlcad Twingy (n=justin@74.92.144.217) |
| 23:33.55 | *** join/#brlcad ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 23:33.55 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || http://fisheye1.cenqua.com/browse/brlcad/brlcad || 7.10 is now released! .. e-mail announcements will follow posting of binary distributions |
| 23:35.56 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/libbu/avs.c: |
| 23:35.57 | CIA-4 | BRL-CAD: first stab at fixing a memory leak on attribute names -- appears to be during a |
| 23:35.57 | CIA-4 | BRL-CAD: merge, where adding to an existing doesn't free the existing name (when a new |
| 23:35.57 | CIA-4 | BRL-CAD: one is provided). clean up the avs printing summary and doxygen comments too. |
| 23:43.09 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/conv/asc2g.c: ws |
| 23:46.41 | *** join/#brlcad ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 23:46.41 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || http://fisheye1.cenqua.com/browse/brlcad/brlcad || 7.10 is now released! .. e-mail announcements will follow posting of binary distributions |
| 23:54.42 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/conv/asc2g.c: close/free file pointers and buffers on shutdown |
| 23:55.36 | *** join/#brlcad MinstrelGypsy (n=mario_du@bas2-sudbury98-1096601318.dsl.bell.ca) |
| 23:56.44 | MinstrelGypsy | tclUnixTime.c in tcl/unix line 609, shouldn't that be _gettimeofday? |