| 00:00.09 | brlcad | I can't say I've really had an issue with emacs undo, it's pretty nifty |
| 00:00.37 | brlcad | your example didn't make much since, you never undid a to redo a |
| 00:00.57 | ``Erik | oh, well, vim records undo trees, not undo lists |
| 00:01.06 | ``Erik | even if my example missed an undo |
| 00:01.46 | ``Erik | "redo b" |
| 00:01.47 | ``Erik | there |
| 00:01.58 | brlcad | not clear to me what that really means in the context of editing files |
| 00:02.41 | ``Erik | if you undo and change something else, either you throw away the 'redo' capability, or you require a tree of changes... vim changed to hold the tree |
| 00:02.53 | brlcad | that exact same list of actions is what emacs would do too |
| 00:04.25 | brlcad | again the trick being that an undo really just adds a new edit event which can itself still be undone |
| 00:04.59 | brlcad | basically lets you play back any/every action all the way back to the last save |
| 00:05.20 | brlcad | including undos and undos of undos and whatever intermediate actions |
| 00:05.24 | ``Erik | hm, I'm not good enough with emacs to have figured out how to use undo well |
| 00:05.44 | ``Erik | ^xu is what I used, I think |
| 00:05.44 | *** join/#brlcad R0b0t1 (n=Enigma@unaffiliated/r0b0t1) |
| 00:05.45 | brlcad | ctrl+shift+_ |
| 00:05.56 | ``Erik | or, uh, C-x u |
| 00:06.22 | brlcad | that's redo |
| 00:07.32 | brlcad | will play the actions back forward again (so say you undid every single action, you could redo everything to get back to where you were regardless of intermediate steps) |
| 00:08.05 | brlcad | pretty darn nifty |
| 00:09.08 | brlcad | sounds maybe like vim caught up with the same feature (but perhaps via diff implementation mechanism) |
| 00:10.07 | ``Erik | hrm, d'no *shrug* not versed enough in emacs to honestly compare 'em, even though I spent most of the day in emacs |
| 00:10.51 | ``Erik | my rough guess would be that understanding undo as a queued event is one thing, but still a lot more linear than going tree based |
| 00:11.57 | brlcad | wow, GSL is about a quarter million lines of code |
| 00:12.32 | ``Erik | http://www.gnu.org/software/gsl/ ? |
| 00:13.41 | brlcad | yep |
| 00:14.29 | brlcad | using their root solving to validate a change to our root solver |
| 00:15.04 | brlcad | last I compared, we get the same results as expected |
| 00:15.18 | brlcad | we were just a solid order of magnitude faster |
| 00:15.41 | ``Erik | what about mathematica or maple or something? |
| 00:15.54 | ``Erik | or writing one up in an absolute system, like scheme or haskell? |
| 00:16.03 | brlcad | what? |
| 00:16.05 | brlcad | what for? |
| 00:16.06 | brlcad | :) |
| 00:16.14 | brlcad | go for it |
| 00:16.18 | ``Erik | to see how bad the floating point fuzz hurts us |
| 00:16.33 | brlcad | i can't imagine mathematica being faster on basic root solving |
| 00:16.39 | ``Erik | no, not faster |
| 00:16.50 | ``Erik | but "more correct", perhaps |
| 00:17.07 | brlcad | though this is at a raw API level I was comparing -- how long to get roots for polynomial of various degree |
| 00:17.41 | brlcad | that was the diff, gsl would do arbitrary degree -- we were compile-time limited to a fixed degree |
| 00:17.53 | ``Erik | ah, aight :) |
| 00:18.07 | ``Erik | I might have some old code from college for root solving via various techniques |
| 00:18.10 | brlcad | i compared about a half-dozen solvers I found around the net |
| 00:18.20 | brlcad | some academic, few packages like gsl |
| 00:18.26 | ``Erik | I vagually recall writing a newtonian one in haskell that was ass slow, the scheme version was a few orders of magnitude faster |
| 00:18.35 | ``Erik | like, finished in minutes instead of not having an answer over the weekend |
| 00:18.48 | brlcad | I thought we were really slow, which is why I was looking |
| 00:18.50 | ``Erik | but single root solver, not general |
| 00:18.59 | brlcad | but turned out that we beat everything I could find and tested against |
| 00:19.19 | brlcad | one other implementation had about the same performance, within a couple percent |
| 00:19.34 | ``Erik | see, that's why I'm asking about correctness... it's easy to generate fast wrong answers... :D |
| 00:19.56 | brlcad | they were all correct answers |
| 00:20.08 | ``Erik | um, like the xkcd commit about random number generators... /* proven correct */ int random() { return 4; } |
| 00:20.09 | brlcad | or at least everyone produced the same results :P |
| 00:20.11 | ``Erik | :) |
| 00:20.15 | ``Erik | within? |
| 00:20.28 | brlcad | within floating point tol |
| 00:20.30 | brlcad | double tol |
| 00:20.35 | ``Erik | okie |
| 00:20.38 | ``Erik | cool beans |
| 00:20.40 | brlcad | close enough |
| 00:21.13 | brlcad | I'm sure there's still plenty of edge case or more extreme polynomials that our simple one can't solve |
| 00:21.53 | ``Erik | I did g-nmg on a file that produces a .g which crashes on dirbuild, any insight? |
| 00:22.13 | brlcad | insight #1: fix it |
| 00:22.20 | brlcad | insight #2: don't do that |
| 00:22.24 | ``Erik | hehehe |
| 00:22.53 | ``Erik | g-nmg -o file-nmg.g file.g toplevel <-- gives a fruity .g file! :D |
| 00:22.54 | brlcad | is full of insightful comments |
| 00:23.25 | brlcad | did g-nmg say it went to completion? |
| 00:23.29 | ``Erik | yup |
| 00:23.35 | ``Erik | it was a big file |
| 00:23.56 | ``Erik | was to come up with a faster loading demo for isst, so you know the file |
| 00:24.19 | ``Erik | I did manage up a 200k face m35 for demoing |
| 00:24.24 | brlcad | my guess would be that some intermediate object failed, but is still referenced -- and dirbuild has some bad juju that assumes the missing object should be there and bites the dust |
| 00:24.54 | brlcad | definitely something to fix in dirbuild |
| 00:25.27 | ``Erik | still at the office? I can tell ya where the criminal .g file is |
| 00:25.45 | ``Erik | can also point you to a .g file that flips out because every effin' matrix is all 0's heh |
| 00:25.58 | brlcad | don't wanna look at it at the moment, or i'll be here till midnight |
| 00:26.07 | ``Erik | or I can bitch at ya for those sun machines |
| 00:26.18 | brlcad | matrix of 0's is probably endian |
| 00:26.30 | brlcad | dirbuild succeeds, but all the floats are hozerd |
| 00:26.31 | ``Erik | v5 db |
| 00:26.46 | brlcad | a v5 from an upgraded v4, upgraded on the wrong plat |
| 00:27.03 | ``Erik | *shrug* perhaps |
| 00:27.13 | ``Erik | trying to remember where I pulled it up last |
| 00:27.39 | ``Erik | it's a simple r/c plane, small model |
| 00:29.50 | brlcad | a good bit of cleanup on the step code.. cut the warnings at least in half I think |
| 00:30.02 | ``Erik | depends on the compiler |
| 00:30.05 | brlcad | wonders what all the "will never be executed" warnings are all about |
| 00:30.18 | ``Erik | ExpDir.h was the majority of warnings for my bsd build |
| 00:31.38 | brlcad | looks like most of those execution warnings are on object constructors.. |
| 00:33.32 | brlcad | tries a sick test |
| 00:34.43 | brlcad | wow, he already did |
| 00:35.51 | brlcad | ~seen madant |
| 00:35.52 | ibot | madant <i=cb7baf0f@gateway/web/freenode/x-a32eed164597bd06> was last seen on IRC in channel #brlcad, 179d 4h 13m 58s ago, saying: 'nothing more disastrous than non-cooperative softwares ;)'. |
| 00:35.59 | brlcad | ~seen homovulgaris |
| 00:36.00 | ibot | homovulgaris <n=d@117.196.131.215> was last seen on IRC in channel #brlcad, 505d 20h 43m 29s ago, saying: 'sean, on a scale of 1 to 10 how much trouble would one face when trying to make a .deb package for brlcad ?'. |
| 00:36.05 | brlcad | hrmph |
| 00:36.33 | brlcad | he really did a nice piece of work fixing our root solver parameters |
| 00:36.44 | brlcad | they were all in reverse order, he flipped them right |
| 00:37.02 | brlcad | and he wrote a little test harness that compares to gsl results |
| 00:37.08 | brlcad | showing the diff |
| 00:38.36 | ``Erik | and lets home gsl isn't flawed O.o :D |
| 00:38.39 | brlcad | http://paste.bzflag.bz/m4050e32c |
| 01:19.23 | brlcad | looks like gsl has improved, only about 5% slower now (very consistently) |
| 01:20.19 | ``Erik | <PROTECTED> |
| 01:21.08 | ``Erik | heheheheheheheh http://www.youtube.com/watch?v=6bs_Lk5iQzc :D always makes me smile |
| 01:53.58 | starseeker | brlcad: that mean the patch with coefficient changes is good to go in? |
| 01:54.24 | brlcad | starseeker: it looks like it, still testing |
| 01:54.30 | starseeker | sweeet |
| 01:54.32 | brlcad | i'll commit, it needed a few minor tweaks |
| 01:54.38 | starseeker | cool |
| 01:54.40 | brlcad | but in all, nice work |
| 01:54.53 | starseeker | is stuck in windoze land anyhow til next year... |
| 01:55.15 | brlcad | the gsl comparison is the best part, but alas .. :) |
| 01:55.21 | brlcad | maybe keep the test harness code as example |
| 01:55.53 | starseeker | nods - maybe have it as an optional part of the regression testing? |
| 01:56.15 | brlcad | the testing to link against gsl is too peripheral |
| 01:56.19 | starseeker | or even some special "make rootcompare" test? |
| 01:56.21 | starseeker | ah |
| 01:57.13 | brlcad | unless... |
| 01:57.29 | brlcad | i make it an option in our solver itself to use gsl as an alternative impl |
| 01:57.37 | brlcad | could be interesting |
| 01:57.41 | starseeker | hmm - that could be cool |
| 01:57.48 | ``Erik | compile time? or? |
| 01:57.59 | brlcad | yep |
| 01:58.09 | ``Erik | 'k, branches are insanely expensive on modern x86 |
| 01:58.33 | ``Erik | we're at what, twenty something deep on the pipeline? |
| 01:59.09 | brlcad | starseeker: did you have a good tgc failure test case handy? |
| 01:59.24 | brlcad | i know there's tire's eto's, they could be tested |
| 02:00.06 | brlcad | lots of complete episodes .. :) |
| 02:00.56 | ``Erik | do we have an m35 with 'good' tires on it? I did a g-nmg on it with 200k triangles in the output, loads up fast in isst and looks pretty decent, but the tires look odd |
| 02:01.39 | ``Erik | and tires should give me lots of triangles in a cheap kd-tree :D |
| 02:02.25 | starseeker | brlcad: havoc has that tgc that's giving rtgl fits, but I didn't isolate it yet |
| 02:03.10 | starseeker | ``Erik: you probably don't want to tesselate one with tire tool tires - iirc the last time we tesselated those it was... um... interesting |
| 02:03.54 | ``Erik | interesting how? I'd imagine it'd be brutal on the nmg csg shtuff, but the results should be clean |
| 02:03.58 | starseeker | brlcad eventually got one when we turned the tread off |
| 02:04.16 | starseeker | took quite a while |
| 02:04.22 | ``Erik | ah, hrm |
| 02:04.56 | starseeker | it's a torture test for the tesselatorer |
| 02:05.19 | ``Erik | damn |
| 02:05.28 | ``Erik | :/ I might have to spend time on tha tmarching cubes crap |
| 02:06.33 | starseeker | extruded sketches interesected with combinations based on etos... |
| 02:07.39 | ``Erik | sounds like a great test case for bug fixing |
| 02:07.52 | ``Erik | is still waiting for brlcad to give him obj files to test obj-g and make that work |
| 02:08.08 | starseeker | can blender generate them? |
| 02:08.30 | starseeker | or wings3d? |
| 02:09.16 | ``Erik | obj files? I'd imagine so... I have a test set, but I'm only going to finish up a subset, brlcad made it sound like he had 'real world' models he wanted converted, so I was hoping to see what subset of functionality from that |
| 02:10.58 | *** join/#brlcad Nohla (n=jesica@168.226.178.49) |
| 02:13.07 | starseeker | should fix that geometry duplication in the tire tool output - that might be making the tesselation unduly difficult |
| 02:16.33 | *** join/#brlcad Nohla (n=jesica@168.226.178.49) |
| 02:33.12 | starseeker | wishes he had that graphviz visualization of the tire tree... |
| 02:35.55 | starseeker | ``Erik: how come the tire command on the bz install of BRL-CAD can't find libtcl? |
| 02:40.53 | ``Erik | hte old bz, or the new bz? |
| 02:40.57 | starseeker | old |
| 02:41.07 | starseeker | ``Erik: how does one get to the new bz? |
| 02:41.12 | ``Erik | cuz it's an ancient piece of shit that needs to be killed? |
| 02:41.23 | starseeker | ``Erik: ah |
| 02:41.42 | ``Erik | no, seriously, it was out of date over 5 years ago |
| 02:44.49 | Nohla | starseeker hello, little prince :) |
| 02:44.56 | starseeker | hola! |
| 02:45.52 | starseeker | note to self - REALLY need to make a smaller version of the tire poster... |
| 02:46.02 | starseeker | redo in inkscape if I have to... |
| 02:46.59 | starseeker | or see if the version of ghostscript on my gentoo box has learned any new tricks... |
| 02:48.44 | ``Erik | you have the photoshop version? it can be saved as a pdf in 'preview' |
| 02:49.00 | starseeker | oh, have pdf |
| 02:49.19 | starseeker | Apple's desktop publishing software was the original generator |
| 02:49.55 | ``Erik | pdf is pretty squished... |
| 02:50.04 | starseeker | probably had images too big |
| 02:57.01 | ``Erik | "save as png"? |
| 02:58.22 | ``Erik | grumbles as his lisp markov 'disassociated text' has some funky issues with new nodes not being created quite right |
| 02:59.43 | ``Erik | it was all disgustingly easy until the recursive call to generate the dictionary ate a wrong type, and now it's hard figure out why *grumble* is assoc not doing what the docs say? |
| 03:00.46 | starseeker | who's fault is it? sbcl? |
| 03:00.53 | ``Erik | well |
| 03:01.40 | ``Erik | (setf (cadr d) (acons (pair l) '(()()) (cadr d))) |
| 03:02.22 | ``Erik | (add-dictionary-word (assoc (pair l) (cadr d)) l 0) |
| 03:02.35 | ``Erik | that assoc keeps coming up with nil, when it shoulda just been defined |
| 03:03.28 | ``Erik | using sbcl, yes... but assoc should just work, I'm not sure if I'm feeding it wrong |
| 03:04.18 | ``Erik | I'm sure I'm making a stupid mistake, but I've thrashed all over :/ |
| 03:04.43 | *** join/#brlcad Nohla (n=jesica@168.226.178.49) |
| 03:05.32 | starseeker | ``Erik: I take it #lisp would just be insulting? |
| 03:06.22 | ``Erik | I'm not feeling up to snuff to defend myself at the moment... I don't think I've put enough time/thought to it |
| 03:06.42 | ``Erik | <-- notes that he's grumbling here, not asking for help :D |
| 03:07.56 | ``Erik | there's a lisp weirdness where something like '(() ()) comes up as '() in some cases |
| 03:08.07 | ``Erik | I might just need to use a dummy symbol to expand teh lists |
| 03:09.07 | starseeker | ah |
| 03:09.12 | ``Erik | <-- is fighting emacs as much as lisp... gave up on gui emacs and started using -nw this afternoon, coudln't get a nice black background |
| 03:09.50 | ``Erik | y'know, when I wasn't reading webcomics or yelling at indianlarry for breaking teh build :) |
| 03:10.45 | ``Erik | (it's on my idp, damnit.) |
| 03:11.12 | starseeker | problem lives somewhere around tire.c:1371 |
| 03:11.21 | starseeker | tire-upper-sides definitions |
| 03:14.43 | ``Erik | with the duplicated geometry? |
| 03:15.04 | starseeker | yes |
| 03:15.36 | starseeker | getting three instances of EllClip when there appear to be only two mk_addmember calls |
| 03:15.41 | ``Erik | starseeker: do any of your cats have an object they're obsessed with dragging around? |
| 03:17.40 | starseeker | she likes her blue sparkly ball, but she's not terribly obsessed with it |
| 03:17.55 | starseeker | usually doesn't carry anything with her |
| 03:18.29 | ``Erik | ah, I have a furry cat fishing pole, the boy needs to drag his 'kill' all over the place |
| 03:18.31 | starseeker | ``Erik: when's the last update to BRL-CAD on the new server? |
| 03:18.58 | starseeker | can't gdb this sucker very well... |
| 03:19.04 | ``Erik | um, /usr/brlcad is an svn from a week or two ago, /usr/local/brlcad is like 7.14.smomething |
| 03:19.13 | starseeker | hmm |
| 03:19.14 | starseeker | ok |
| 03:19.38 | ``Erik | feel free to build your own :) |
| 03:19.51 | starseeker | ``Erik: yeah, she'll "kill" her catnip pillow sometimes |
| 03:20.26 | ``Erik | I keep putting this thing in the living room downstairs, it keeps ending up in my bed O.o |
| 03:20.33 | starseeker | hehe |
| 03:21.17 | ``Erik | updating /usr/brlcad/HEAD with a fresh checkout now |
| 03:21.32 | starseeker | will have to step through this to figure out why the frap the boolean logic is odd... |
| 03:22.39 | starseeker | may have to wait til I have better setup (sorry ``Erik :/) |
| 03:23.14 | ``Erik | *shrug* ain't no skin off my nose |
| 03:23.15 | starseeker | I'm not real sure why the standard m35 tires are hozed |
| 03:23.21 | starseeker | they shouldn't be |
| 03:23.22 | *** join/#brlcad Nohla (n=jesica@168.226.178.49) |
| 03:23.24 | ``Erik | hozed? |
| 03:23.34 | ``Erik | the m35 tires convert just fine, they're just ... boring |
| 03:23.41 | starseeker | when Nick did his rtgl stuff, one side of the m35 tires didn't raytrace right |
| 03:23.51 | starseeker | might not actually be the fault of the tires though |
| 03:23.57 | starseeker | oh, OK ;-) |
| 03:24.33 | starseeker | did do an m35 with fancy tires, but I think the "correct" size looked kinda puny |
| 03:24.40 | ``Erik | I was figuring that procdb tires would look a lot better and provide a lot mroe triangles without any signicant deepening of the kd-tree |
| 03:24.49 | starseeker | bigger ones overlaped, which I could ignore for screenshots but isst would kinda care |
| 03:25.07 | ``Erik | and I'd recalled seeing and m35 with 2 tires replaces, so'z I was wondering if there was one with all 11 or whatever redone |
| 03:25.23 | starseeker | true, although the tires have some tree depth (4 or 5 levels, forget offhand) |
| 03:25.34 | ``Erik | (2+8+1, right?) |
| 03:25.51 | starseeker | yeah, I had done it once, but I don't know if I ever got it "right" in the overlap free sense |
| 03:25.53 | ``Erik | I doubt we're talking the same trees |
| 03:26.03 | starseeker | probably not |
| 03:26.12 | starseeker | is talkin csg tree |
| 03:26.22 | ``Erik | libtie kdtree's would fluff maybe a dozen or two in the tire areas |
| 03:26.33 | starseeker | sh |
| 03:26.35 | starseeker | er ah |
| 03:26.47 | ``Erik | kd-tree's are axis aligned binary splitting plane trees |
| 03:27.23 | starseeker | I think the "right" way to do that is to generate tires without wheels and then sub in for just the "rubber" part - should be able to minimize/avoid overlaps that way |
| 03:27.39 | starseeker | the m35 wheel is rather different from the tire tool's default wheel |
| 03:27.46 | ``Erik | the wheels you defined are very not military style |
| 03:28.06 | starseeker | yeah, was just doing something "quick and dirty" |
| 03:28.27 | ``Erik | heh, too much "pretty", not enough "quick and dirty" :D |
| 03:28.39 | starseeker | lots of fun that could be had with that tool, given time |
| 03:29.04 | ``Erik | I can give you a pretty correct rim for a stryker or hummer, they're utterly boring |
| 03:29.17 | ``Erik | the most excitement is hidden by the rubber on 'em |
| 03:29.27 | starseeker | ``Erik: <snort> the whole thing was an exercise in "doing a proc-db", just like the pen model was "learning mged" - establishing early on my obsession with unnecessary detail ;-) |
| 03:29.36 | ``Erik | y'know |
| 03:29.47 | ``Erik | one of these days, I'll do something like the pen exercise O.o |
| 03:30.04 | starseeker | ``Erik: yeah, for most real applications tire without wheel is probably how that will get used - most people have purty wheels of their own |
| 03:30.25 | ``Erik | <-- came in and was sat down with a machine, an os disc and a source disc... told "port this to fbsd 5.1, then convert it to autotools" |
| 03:30.41 | starseeker | ``Erik: what, you learn mged? Let me know when you do so I can pre-book skating tickets for hell |
| 03:30.45 | ``Erik | never had a "learn how to model" phase |
| 03:30.58 | starseeker | or you could wait til MGED 2.0 I suppose... |
| 03:31.16 | ``Erik | no, seriously, I wasn't allowed on the network, so I got a throw-away machine, os disc and a cd with the source on it... |
| 03:31.23 | ``Erik | and parked in a room |
| 03:31.29 | starseeker | ah :-) |
| 03:31.29 | ``Erik | next to mike |
| 03:32.03 | ``Erik | first day, "here's a mchine, here's an OS disc, here's the source... port it." |
| 03:32.13 | starseeker | hehe |
| 03:32.23 | starseeker | yeah, I came in a lot dumber :-P |
| 03:32.30 | starseeker | had to get up to speed |
| 03:33.01 | ``Erik | I had bsd chops and auto* chops, but I did an awful lot of asking people to test what I'd done to see if I'd done it right... |
| 03:33.09 | ``Erik | e.e. was one of my primary victems |
| 03:33.25 | ``Erik | "bubba" |
| 03:33.49 | ``Erik | and you came in knowing a lot more about BRL-CAD than I did |
| 03:34.16 | ``Erik | they were nice to you :D |
| 03:34.17 | starseeker | well, the coding/automake chops let you start doing "useful work" pronto |
| 03:34.29 | starseeker | heh |
| 03:35.18 | ``Erik | auto* was a "possible way to go" at the time, not a guaranteed thing, it was very much cake |
| 03:35.44 | starseeker | well, it sure worked out in the end :-) |
| 03:37.46 | ``Erik | I put a lot of work into it, then brlcad went and did horrible horrible things to my baby |
| 03:37.57 | brlcad | heh |
| 03:37.59 | ``Erik | and now it's a frankenstein monster :D |
| 03:38.05 | brlcad | now it's complete |
| 03:38.29 | starseeker | so the docbook stuff is just the icing on the cake in frankensteinian terms? :-P |
| 03:38.36 | ``Erik | um, I think the tag exists in svn, something liek AUTOCONF_200311XX |
| 03:39.15 | ``Erik | so when's the old crud on brlcad.org migrating to crit, brlcad? :) |
| 03:40.30 | starseeker | brlcad: I haven't had a chance to test it - did adding the mkdir commands to the docbook build rules avoid that issue on highly parallel docbook builds? |
| 03:41.20 | brlcad | starseeker: dunno |
| 03:41.33 | starseeker | ah, k - have to check when I get back in |
| 03:41.50 | ``Erik | 20031203 |
| 03:42.02 | ``Erik | wow,w as much simpler back then, configure.in was 225 lines |
| 03:42.02 | starseeker | also needs to check if the 3dm-g tool got busted by the opennurbs upgrade... |
| 03:44.41 | starseeker | also need to figure out how to enable the tk framebuffer without killing mged... still not sure why that's happening |
| 03:45.10 | starseeker | kinda hard to figure out how complete it is when it takes out the whole gui :-( |
| 03:45.46 | starseeker | notes more SIGGRAPH spam has arrived... |
| 04:12.05 | starseeker | hrm - build fails |
| 04:13.27 | starseeker | GlobalUnitAssignedContext.cpp:27 in SdaiCONFIG_CONTROL_DESIGN.h:2534 error expected "," or "..." before '&' token |
| 04:14.06 | starseeker | calls it a night |
| 04:18.36 | *** join/#brlcad alex_jon1 (n=alex_jon@81.196.65.201) |
| 04:50.51 | brlcad | no /me doesn't get that error |
| 07:22.27 | *** join/#brlcad louipc (n=louipc@archlinux/trusteduser/louipc) |
| 11:04.01 | *** join/#brlcad mafm (n=mafm@162.Red-81-32-97.dynamicIP.rima-tde.net) |
| 11:21.06 | *** join/#brlcad mafm (n=mafm@162.Red-81-32-97.dynamicIP.rima-tde.net) |
| 11:24.41 | *** join/#brlcad R0b0t1 (n=Enigma@unaffiliated/r0b0t1) |
| 11:42.30 | ``Erik | dang, yet more tkhtml3 breakage |
| 12:22.01 | ``Erik | (apparently, tclsh isn't always named tclsh, sometimes it's tclsh85) |
| 13:07.15 | ``Erik | ßs hrm http://www.collegehumor.com/video:1926917 |
| 14:41.41 | d-lo_ | ``Erik: Sent you an email at your gmail and bzflag.bz accounts. |
| 14:42.18 | ``Erik | thanks |
| 15:01.38 | CIA-38 | BRL-CAD: 03brlcad * r37067 10/brlcad/trunk/src/librt/primitives/bot/ (bot.c g_bot_include.c): make RT_BOT_UNORIENTED_NORM() macro take the in/out 'param' as a bool for indicating exit instead of toggling on positive/negative. also nix the RT_ prefix as this is not public api. |
| 15:41.35 | brlcad | haha... http://www.collegehumor.com/video:1735402 (nsfw, but just because of language) |
| 15:42.03 | brlcad | that's classic b'more |
| 16:12.10 | *** join/#brlcad mafm (n=mafm@162.Red-81-32-97.dynamicIP.rima-tde.net) |
| 16:13.56 | ``Erik | oh, the car commercial, yeh :D |
| 16:18.54 | *** join/#brlcad mafm_ (n=mafm@63.Red-83-45-252.dynamicIP.rima-tde.net) |
| 16:21.46 | CIA-38 | BRL-CAD: 03indianlarry * r37068 10/brlcad/trunk/src/ (7 files in 2 dirs): more warning quellage, also pushed back mods to fedex_plus generated code |
| 16:47.15 | d-lo_ | that whole FEDEX thing in the step code throws me off every time I see it. lol |
| 17:10.08 | *** join/#brlcad akafubu (n=akafubu@unaffiliated/akafubu) |
| 18:18.15 | CIA-38 | BRL-CAD: 03brlcad * r37069 10/brlcad/trunk/src/librt/primitives/bot/bot.c: quell verbose warnings. there were two exact floating point comparisons in here, so potential behavior change (i.e., hopefully more robust/consistent). |
| 18:52.46 | *** join/#brlcad R0b0t1 (n=Enigma@unaffiliated/r0b0t1) |
| 19:02.33 | ``Erik | heh, I used to work at FedEx, so imagine what it does to me :D |
| 19:07.31 | CIA-38 | BRL-CAD: 03brlcad * r37070 10/brlcad/trunk/src/librt/primitives/ (6 files in 3 dirs): more slews of quellage and cleanup, dead code elimination, exact floating point comparisons, and param checking |
| 19:24.15 | brlcad | mm, the problem was exactly what bob and I speculated last night |
| 19:24.29 | brlcad | looks like windows is being harsh on the tclIndx generation, at least more strict |
| 19:24.36 | brlcad | and not indexing some files |
| 19:25.21 | CIA-38 | BRL-CAD: 03erikgreenwald * r37071 10/brlcad/trunk/misc/win32-msvc8/ (adrt/adrt.vcproj brlcad/brlcad.sln): add adrt to brlcad project |
| 19:31.03 | ``Erik | Yes. I broke it. Real bad. |
| 19:32.34 | ``Erik | and msvc has been 'not responding' for like fifteen minutes, so *shrug* |
| 19:59.39 | CIA-38 | BRL-CAD: 03brlcad * r37072 10/brlcad/trunk/BUGS: too vague a bug, msvc build files have changed drastically since then anyways |
| 20:02.58 | CIA-38 | BRL-CAD: 03brlcad * r37073 10/brlcad/trunk/BUGS: |
| 20:02.58 | CIA-38 | BRL-CAD: mged on windows is not generating the tclindex files correctly. could be a |
| 20:02.58 | CIA-38 | BRL-CAD: simple parsing error or some other problem in the code bug a variety of files |
| 20:02.58 | CIA-38 | BRL-CAD: aren't getting indexed (notably the pattern tool and sketch editor but also some |
| 20:02.58 | CIA-38 | BRL-CAD: simple commands like bot and per_line. |
| 20:16.16 | CIA-38 | BRL-CAD: 03brlcad * r37074 10/brlcad/trunk/src/librt/primitives/bspline/ (10 files): quell all of the old nurbs code verbse warnings safe for one bogus unreachable code warning on a goto label. lots and lots of exact floating point comparisons.. should be more robust now even if it is going away rsn. |
| 20:17.45 | CIA-38 | BRL-CAD: 03brlcad * r37075 10/brlcad/trunk/BUGS: reported that fbclear is not working on windows build of mged. |
| 20:31.11 | ``Erik | òò/cl |
| 20:32.34 | CIA-38 | BRL-CAD: 03brlcad * r37076 10/brlcad/trunk/include/opennurbs_ext.h: unused quellage |
| 20:34.04 | CIA-38 | BRL-CAD: 03brlcad * r37077 10/brlcad/trunk/ (3 files in 2 dirs): add a bu_list parameter so we can make rt_nurb_intersect get called recursively when surfaces are split. this is an alternative to the former goto method and quells an unreachable warning. |
| 20:34.33 | *** join/#brlcad Ralith (n=ralith@216.162.199.202) |
| 20:44.15 | CIA-38 | BRL-CAD: 03brlcad * r37078 10/brlcad/trunk/src/librt/primitives/dsp/dsp.c: quell verbose dsp warnings. lots and lots of exact floating point comparisons to zero and one changed to within tol tests. |
| 20:44.49 | CIA-38 | BRL-CAD: 03brlcad * r37079 10/brlcad/trunk/src/librt/primitives/ (cline/cline.c dsp/dsp_brep.cpp): more param and unused var quelling, s/log/logstr/ too |
| 20:46.36 | ``Erik | hehehe http://video.google.com/videoplay?docid=36099539665548298 |
| 20:55.08 | brlcad | that's an oldie |
| 21:08.22 | ``Erik | src/librt/primitives/nmg/nmg_rt_isect.c:1936: error: too few arguments to function 'rt_nurb_intersect' |
| 21:08.36 | ``Erik | forget to commit something? |
| 21:18.51 | *** join/#brlcad akafubu (n=akafubu@unaffiliated/akafubu) |
| 21:26.03 | *** join/#brlcad SWPadnos (n=Me@dsl107.esjtvtli.sover.net) |
| 21:29.20 | CIA-38 | BRL-CAD: 03brlcad * r37080 10/brlcad/trunk/src/librt/primitives/nmg/nmg_rt_isect.c: blechk! nmg cross-pollution with the old nurbs code.. rt_nurb_intersect() had another param added. |
| 21:29.33 | brlcad | not reading your commit diffs, eh? |
| 21:29.51 | ``Erik | mine? huh? |
| 21:30.42 | brlcad | nvr mind |
| 21:40.40 | CIA-38 | BRL-CAD: 03erikgreenwald * r37081 10/brlcad/trunk/src/conv/iges/trimsurf.c: add missing parameter |
| 21:41.55 | brlcad | wow |
| 21:41.56 | brlcad | that's just wrong |
| 21:42.10 | ``Erik | what, did I break something horrible? |
| 21:42.12 | brlcad | iges converter is actually shooting rays at nurbs |
| 21:42.23 | ``Erik | oh, heh |
| 21:42.40 | brlcad | what you did is fine |
| 21:42.45 | ``Erik | <-- just did a monkey mimic of your patch to that file, didn't bother, y'know, looking |
| 21:42.47 | brlcad | the fact that the iges converter is doing that.... |
| 21:43.02 | brlcad | that any code is calling that nurb internal routine |
| 21:43.07 | ``Erik | so wasn't sure if I screwed that up :) |
| 21:43.12 | brlcad | nah |
| 21:43.51 | brlcad | i'm still on strict build, so can't build through to get to any latent callers |
| 21:44.27 | ``Erik | so build out of dir and have two builds, one strict and one not :D |
| 21:48.15 | brlcad | this is already so utterly painfully tedious as it is to make it twice as painful |
| 21:48.41 | brlcad | all to prevent the 1/100 chance that there's some triviality later in the build |
| 21:50.59 | brlcad | it's also not like anyone else is jumping in to help get these all taken care of more quickly either :P |
| 22:00.40 | ``Erik | yeah, damn shame about that :D |
| 22:03.49 | ``Erik | rt_bot_tess() *sigh* why didn't I think of that earlier? |
| 22:06.57 | CIA-38 | BRL-CAD: 03brlcad * r37082 10/brlcad/trunk/src/librt/primitives/ (8 files in 6 dirs): mas quellerismos, floating point comparisons, unused vars/params |
| 22:09.51 | CIA-38 | BRL-CAD: 03brlcad * r37083 10/brlcad/trunk/src/librt/primitives/ (extrude/extrude.c half/half.c table.c): s/free/release/ to avoid shadowing the function |
| 22:53.44 | *** join/#brlcad Nohla (n=jesica@201.255.221.182) |
| 23:39.00 | brlcad | mm.. this should be interesting |
| 23:39.06 | brlcad | floating point comparisons in nmg |
| 23:39.28 | ``Erik | heh, um, have fun with that :D *duck* |
| 23:39.45 | ``Erik | nmg's already raytrace arse slow |
| 23:40.05 | brlcad | there are in boolean evaluation, should help |
| 23:40.37 | ``Erik | y'know, I"ve seen geometry errors thrown on nmg tesselation... |
| 23:40.45 | ``Erik | starting with facet type nmg's... |
| 23:41.16 | ``Erik | I'm not sure if I'm willing to commit the time to chase that just yet, I have other pressing issues to worry about |
| 23:41.38 | ``Erik | but, ... yeah, wtfage :) |
| 23:41.51 | brlcad | geometry errors on nmg tess are normal |
| 23:42.02 | brlcad | part of it guessing which paths to take |
| 23:42.22 | ``Erik | soooo, nmg->nmg is expected to fail? |
| 23:42.24 | brlcad | non-facet nmgs are no longer relevant |
| 23:42.36 | ``Erik | right, but still listed as a flag |
| 23:42.41 | ``Erik | and used in some bits |
| 23:42.42 | brlcad | nmg->nmg could fail depending how the original nmg was created |
| 23:42.47 | ``Erik | g-nmg |
| 23:42.50 | brlcad | if it wasn't validated |
| 23:43.17 | brlcad | that's so high level as to be non-sensical to ask .. |
| 23:43.29 | brlcad | too many variables |
| 23:43.32 | brlcad | above regions, below regions |
| 23:43.34 | brlcad | combs, prims |
| 23:43.56 | brlcad | is it combining regions or leaving them as separate shels |
| 23:44.00 | ``Erik | g-nmg likes to convert a region to a single nmg and attach that as a single object under the region |
| 23:44.25 | ``Erik | (EXACTLY what my 'recursive facetizer' card does... can kill that from Ed's wall) |
| 23:44.40 | brlcad | right, but how you were running it changes things too -- you feed it something under a region, for example, or something with a mixed hierarchy where there are paths that are regionless, etc |
| 23:45.06 | ``Erik | hm, saw it on m35 |
| 23:45.21 | brlcad | just saying it's way to vague a notion to say anything meaningful, way too many variables unaccounted for |
| 23:45.30 | brlcad | can't even say it's clean input |
| 23:45.39 | ``Erik | I'll look into it a bit when I'm in the office tomorrie |
| 23:45.53 | brlcad | m35 has problems |
| 23:45.59 | brlcad | whether they matter or not.. dunno |
| 23:46.11 | brlcad | it's gone both ways in the past |
| 23:46.36 | ``Erik | either way, there are issues to be addressed wrt... |
| 23:47.15 | ``Erik | if our reference geometry is screwed up, that's bad... if our tools produce bad output... that's bad :) |
| 23:47.52 | brlcad | I think the problem has to be brought back down to basic test cases, make sure primitives all tessellate (which they won't) .. then 0-level combinations |
| 23:48.36 | ``Erik | I have a well tesselated m35 (10x better than default tolerance), if it's useful... had one going where the side mirrors gave 200k triangles, killed it :/ |
| 23:49.05 | ``Erik | on the up side, these issues are giving me impetus to do the marching cubes thang |
| 23:49.24 | ``Erik | "cuz lee said" kinda deflated me |
| 23:49.31 | brlcad | g-nmg producing a bad nmg just isn't a good measure of anything -- if it did something wrong, could be a simple bug isolated to that specific tool or the tess approach it uses |
| 23:50.10 | ``Erik | the tool itself is too trivial to break |
| 23:50.19 | brlcad | the approach most of those use isn't so hot |
| 23:50.24 | ``Erik | too small to fail :D |
| 23:50.26 | brlcad | I opted for a slightly diff approach for the ef work |
| 23:50.31 | ``Erik | ef? |
| 23:50.31 | brlcad | more robust |
| 23:50.57 | brlcad | endgame |
| 23:51.03 | ``Erik | ah |
| 23:51.30 | ``Erik | I thought our contribution to their federation was more "give us a geometry name and ray, we give you a prtition list"? |
| 23:51.41 | brlcad | nothing reusable code-wise, but just a note that the approach they use now isn't consistent across the exporters |
| 23:51.47 | ``Erik | well, no, they do ogl visualization, ya had to gie them triangels |
| 23:51.54 | ``Erik | had to give them triangles |
| 23:52.04 | CIA-38 | BRL-CAD: 0367.232.178.197 07http://brlcad.org * r2055 10/wiki/MGED_CMD_eac: |
| 23:52.13 | brlcad | not at all, did a geometry exporter -- basically .g to their format |
| 23:52.13 | CIA-38 | BRL-CAD: 0367.232.178.197 07http://brlcad.org * r2056 10/wiki/MGED_CMD_whichair: |
| 23:52.17 | brlcad | which was poly |
| 23:52.31 | ``Erik | ok, so an exporter, not a federation entity |
| 23:59.46 | brlcad | oh wow |
| 23:59.50 | brlcad | the code is still in there... |