00:00.06 | ``Erik | and printf can puke at complex or generated format strings, or if stdout is gimpy, so a nice 'safe' (without allocations or fmt parsing) display to stderr MIGHT be helpful... :D |
00:02.13 | ``Erik | kinda looks like a bsd-ish license, ya gonna post your multics tarball somewhere? or ask 'em to provide one on their site? heh |
00:27.53 | *** part/#brlcad rpaddock (n=bob_padd@c-24-131-109-50.hsd1.oh.comcast.net) |
01:50.36 | starseeker | Heh - If they were going to provide one I'd assume they would have done so already - it's the most obvious thing to do. |
01:51.02 | starseeker | As for posting it somewhere, I would but I'm not sure where I can post it - I don't really have my own website... |
01:53.10 | ``Erik | I d'no |
01:53.12 | ``Erik | it's mit... |
01:53.32 | starseeker | They may not want a lot of nerds randomly downloading 118 megs... |
01:53.33 | ``Erik | some smart people there, but I think almost all of them thnk they're a lot smarter than they are |
01:57.28 | starseeker | Heh - probably. |
02:03.39 | starseeker | Grrrrrrr. What is Gentoo up to on the AMD64 platform with itcl/itk?? |
02:17.19 | ``Erik | hey, uh, brlcad |
02:17.29 | ``Erik | ctrl+opt+cmd+8 |
02:19.04 | poolio | inverted colors iirc? |
02:46.32 | brlcad | yes? |
02:47.57 | brlcad | starseeker: you're welcome to an account on .bz if you want |
02:53.59 | ``Erik | negative, yes, quite sharp of you, hoolio |
02:54.27 | starseeker | brlcad: That would be nice.. |
02:54.49 | ``Erik | speaking of, is that file still accurate? |
02:55.51 | poolio | ``Erik: I only know it cause I'd use it for mischief on others' computers. |
02:57.03 | ``Erik | heh |
02:57.07 | ``Erik | um |
02:57.31 | ``Erik | in '96 or so, I convinced a friend to get an x86 and install linux... he was a mac guy |
02:57.39 | ``Erik | he invited me in and rooted me up to check over his box |
02:57.50 | ``Erik | so I dd's the kernel to the audio out |
02:57.56 | ``Erik | and opened and closed the cdrom a few times |
02:57.59 | ``Erik | freaked him out :D |
02:58.38 | poolio | heh, i was more of a fan of cat, but yeah, that's always fun. |
02:58.54 | poolio | man, 2nd semester in a row Im not getting the programming course I want. this sucks. |
02:59.08 | ``Erik | what course? |
02:59.10 | starseeker | brlcad: Mightn't they put their eyebrows up at a 118M tarball of an ancient OS though? ;-) |
02:59.33 | poolio | Intro to Computer Systems...there are already 110 people on the waitlist (all not freshmen) and the class is only 150 |
02:59.52 | ``Erik | SOUTHPARK TIME |
03:00.06 | ``Erik | aren't you a little past that? |
03:00.24 | poolio | ``Erik: It's low level C, doing stuff like implementing a proxy, rewriting malloc, etc... |
03:00.30 | louipc | what's Intro to Computer Systems? "this is the monitor, this is the keyboard?" |
03:00.31 | poolio | I know most of it but not enough to place out of it |
03:00.32 | ``Erik | oh |
03:00.42 | ``Erik | aight, cool |
03:00.45 | louipc | ah ok :D |
03:00.56 | poolio | Prepares you for the class after that, OS, which is ridiculously hard...you write your own OS :P |
03:01.08 | louipc | damnit Erik |
03:01.10 | ``Erik | I was afraid you meant like "this program is called excdell, you can put numbers in it" type work |
03:01.17 | ``Erik | writing an OS all depends on the hw |
03:01.36 | ``Erik | x86 is horribly ugly with the nasty gdt/idt/isr hacks |
03:01.52 | ``Erik | um, proxy's and malloc are pretty easy |
03:02.16 | poolio | yeah, I mean 213 shouldn't be too bad, but it will be good practice before I get my ass kicked in OS |
03:02.41 | ``Erik | cmu is actually the school I really want ot go to for my grad work |
03:03.06 | ``Erik | of the handful of 'best for cs' schools, it matches my interests most |
03:03.30 | ``Erik | <-- very into os's and languages |
03:04.04 | louipc | do you folks know a good school for electrical engineering? |
03:07.20 | starseeker | ``Erik: How are you going to go there while working on BRL-CAD? |
03:07.54 | ``Erik | in theory, ARL "supports" 'long term training' |
03:08.01 | starseeker | Ah. |
03:08.26 | starseeker | brlcad: Should I send you any info for the .bz account? |
03:08.59 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/src/mged/setup.c: |
03:09.01 | CIA-27 | BRL-CAD: AHA! I had added the final tclcad_auto_path() call... just not to the mged init |
03:09.03 | CIA-27 | BRL-CAD: routine. it was only added to bwish (those two really should syncronize their |
03:09.05 | CIA-27 | BRL-CAD: init routines into tclcad or something). this "should" fix the "gui" error. |
03:09.14 | ``Erik | real name, shell preference, and mebbe ssh pub keys, otherwise he'd have to make you a temp passwd |
03:09.25 | brlcad | starseeker: they? |
03:09.39 | starseeker | Whoever hosts bz and pays for bandwidth |
03:09.45 | brlcad | .bz is not a gov't server |
03:09.52 | brlcad | ah, then 'they' == 'me' :) |
03:09.54 | ``Erik | 'they' is brlcad |
03:09.54 | starseeker | Ah :-) |
03:10.25 | ``Erik | (though I might try to do something with the machine that I might feel obligated to hand brlcad a little scratch for in the future...." |
03:10.44 | brlcad | it's got several names that point to it at this point, the fqdn of the primary ip is bz.bzflag.bz |
03:11.13 | ``Erik | brlcad, did I miss anything in my list of info? |
03:11.14 | brlcad | though it's also my.brlcad.org, ftp.brlcad.org, and a couple dozen other names |
03:11.40 | brlcad | what list of info? |
03:11.50 | ``Erik | real name, shell, ssh pub key... |
03:12.00 | ``Erik | like, 2 minutes ago... |
03:12.12 | ``Erik | fucking attention span of a squirrel on speed |
03:12.14 | brlcad | oh, I missed that (rather, i've only skimmed my logs |
03:12.20 | ``Erik | :D |
03:12.30 | brlcad | ah, there it be after cia |
03:12.51 | brlcad | which, luckily we don't have one of the broken cia bots |
03:13.18 | brlcad | one of them is rather dead and has been for many days now, scanline's been pretty latent in doing anything about it |
03:14.11 | brlcad | yeah, that's almost everything |
03:42.30 | ``Erik | I installed a bunch of stuff, btu I didn't do "setup", which should just be copying files |
03:43.14 | ``Erik | uh |
03:43.23 | ``Erik | 18% faster than the old machine??? |
03:43.47 | brlcad | yeah :/ |
03:43.48 | ``Erik | oohhhh, wait |
03:43.57 | brlcad | cpu |
03:43.59 | ``Erik | it was already a reasonably fast machine |
03:44.05 | ``Erik | 2.4 celery |
03:44.08 | brlcad | yeah, it's not too shabby |
03:44.19 | ``Erik | I dunno what I was thinking it was like a 667mhz |
03:44.24 | brlcad | went from 2.4 celery to 2.4 penguin |
03:46.24 | ``Erik | just a p4 |
03:46.32 | ``Erik | not like ya stepped up to an opteron |
03:47.55 | ``Erik | all the accounts are 'migrated'? sans home dir? |
03:51.39 | brlcad | gubby mint connotation wasn't implied :) |
03:51.45 | ``Erik | heh |
03:51.46 | ``Erik | well |
03:52.04 | ``Erik | penguin ain't a gov't co, they just got real lucky selling some silly-cone |
03:52.13 | brlcad | yep, all accounts work like a charm now .. migrating the filessytems this weekend |
03:53.05 | ``Erik | in I think '99, I wrote an opengl screen saver for a company doing high perofrmance linux stuff |
03:53.34 | ``Erik | I dont' even remembertheir name |
03:54.03 | ``Erik | but they were based in kc, several of my local buddies went up there to work for them, they kinda snuck me the ss dealie |
03:54.38 | ``Erik | wanted their logo displayed over apool of rippling water... I added a couple moving lights influenced by a mutual gravity type algorithm |
03:54.53 | ``Erik | then a couple formed another company in kansas to compete |
03:55.21 | ``Erik | and one of them went and committed suicide :( |
03:55.51 | ``Erik | 'atipa' in kc, |
03:58.09 | starseeker | And here it is, thanks to brlcad - a Multics tarball :-) |
03:58.20 | starseeker | http://my.bzflag.bz/~starseeker/ |
04:01.45 | ``Erik | brlcad, I put our mortal accounts in the wheel group... Idunno who else is on admit, but it might be worth making that disappear |
04:04.56 | brlcad | making what disappear? |
04:05.05 | brlcad | oh, making admin disappear |
04:05.12 | brlcad | yeah, mebbie |
04:05.26 | starseeker | Well, now all we need is a Multics hardware simulator and we can try reliving the glory days on Multics with BRL-CAD ;-) |
04:05.34 | brlcad | though I usually put an impossible password on that and use that when I have to give the ISP a user/pass to fix things |
04:05.49 | starseeker | Hehehe |
04:24.04 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/src/bwish/main.c: match what mged does order-wise |
04:29.56 | ``Erik | I assume that the set of people with access to that file is greater than 2 |
04:41.31 | brlcad | yep |
04:42.14 | brlcad | it'll go through surges of completion, my goal is to be done by the end of this month |
04:42.22 | brlcad | so a bunch this weekend and next |
05:07.07 | poolio | Hmm, could I ask a question re: svn organization? |
05:07.25 | brlcad | go for it |
05:08.05 | brlcad | there's two main common techniques |
05:08.11 | brlcad | s/techniques/structures/ |
05:09.06 | poolio | So I'm working on a project with a bunch of people, and the established method they have is each sub-project has its own branch, and when their code is stable they commit it to trunk. The issue I am having is that it's a pain to download an entirely diferent branch when you want to see their stuff and you end up with like 20 different branches |
05:10.03 | poolio | Is that considered a normal practice? |
05:19.45 | brlcad | there's no universal practice, that's not unheard of in the least |
05:20.22 | brlcad | just rather heavy-process undoubtedly to maintain the trunk as stable as possible |
05:21.00 | CIA-27 | libirc: 03JeffM2501 * r298 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: add echo support |
05:24.18 | ``Erik | um, I know with cvs, the usual behavior I've seen is most hot development happens in head |
05:24.42 | ``Erik | releases are either rbanchend or tagged (big ones branch, sometimes two levels like fbsd) |
05:25.16 | ``Erik | any radical 'subproject' might get a branch, sometimes in a different VCS |
05:25.34 | ``Erik | fbsd has a lot of 'radicals' in perforce, for example |
05:29.27 | brlcad | that's also in part because branches are more of a bitch in cvs than svn, so they're more minimized |
05:30.07 | brlcad | svn isn't great on the branches, but it does do at least improve upon how easy it is to maintain a branch and merge |
05:31.07 | brlcad | but i've heard that same sub-project structure in cvs where each dev maintains a branch |
05:31.38 | brlcad | just a matter of familiarity/comfort with the level of effort, and the stability/integration tradeoffs |
05:32.17 | brlcad | which apparently isn't your cup of tea, but if there are already that many devs working that way, then they're probably quite accustomed to that process (asuming the project is active) |
05:34.22 | brlcad | hm, did just verify that yukonbob's bad pixels are floating point fuzz |
05:38.45 | CIA-27 | libirc: 03JeffM2501 * r299 10/trunk/libirc/TODO: ws |
05:40.03 | ``Erik | if you go branch heavy, EVERYONE needs to be damn good at resolving conflicts |
05:40.17 | ``Erik | otherwise, everyone needs to be good and updating and commiting often |
05:40.23 | brlcad | yeah |
05:40.24 | CIA-27 | libirc: 03JeffM2501 * r300 10/trunk/libirc/TODO: ws2 |
05:41.07 | brlcad | it'd suck for my taste to go to that extent, but I can easily see projects/people that'd go that route for grandios claims |
05:41.28 | CIA-27 | libirc: 03JeffM2501 * r301 10/trunk/libirc/TODO: ws 3 |
05:43.08 | CIA-27 | libirc: 03JeffM2501 * r302 10/trunk/libirc/TODO: still more whitespace |
05:45.07 | poolio | Yeah, it's firmly established, and I'm just a new freshmen so it's not likely that I can make them change. It works for them though, just has been kinda a pain to keep track of and see what the different senior members are working on |
05:45.14 | poolio | And thanks for the replies guys :) |
05:46.41 | CIA-27 | libirc: 03JeffM2501 * r303 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: Add CIA echo capability |
05:49.53 | ``Erik | blah |
05:50.09 | CIA-27 | libirc: 03JeffM2501 * r304 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: only parse the master if we aren't doing an echo |
05:51.58 | CIA-27 | libirc: 03JeffM2501 * r305 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: log the target channel on an echo |
06:04.07 | CIA-27 | libirc: 03JeffM2501 * r306 10/trunk/libirc/examples/stupidBot/readme.txt: mention echos |
06:04.52 | CIA-27 | libirc: 03JeffM2501 * r307 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: allow more then one CIA echo target per input channel and project |
06:17.41 | CIA-27 | libirc: 03JeffM2501 * r309 10/trunk/libirc/src/IRCUserManager.cpp: do a safer erase that won't geek windows on a user part. |
06:39.00 | CIA-27 | libirc: 03JeffM2501 * r311 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: if the line in the channel dosn't have a : then don't even try to parse it for a CIA echo |
07:45.40 | *** join/#brlcad Defcon (n=def@74.17-246-81.adsl-static.isp.belgacom.be) |
09:14.05 | *** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch) |
11:02.06 | *** join/#brlcad elite01 (n=elite01@dslc-082-082-065-178.pools.arcor-ip.net) |
13:21.01 | *** join/#brlcad PrezKennedy (i=Matt@74.86.45.130) |
13:27.54 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/src/mged/qray.c: Mods to use librt's qray routines. This fixes the inability to adjust the format strings and shrinks MGED's code base a tiny bit. Note - after testing on unix the if'ed out code can be removed. |
13:29.42 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/include/raytrace.h: Expose a few more functions in librt. |
14:12.40 | *** join/#brlcad poolio (n=poolio@c-71-236-67-238.hsd1.oh.comcast.net) |
14:20.45 | brlcad | howdy poolio |
14:21.44 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/misc/win32-msvc8/Makefile.am: Added a few more projects. |
14:22.10 | poolio | mornin' |
14:25.41 | ``Erik | yargh, matey |
14:27.30 | poolio | All my classes are slowly but surely filling up and I can't register for another hour |
14:27.36 | poolio | grargh. |
14:31.41 | ``Erik | that's usually the way it goes... seniority gets first stab... |
14:32.03 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/misc/win32-msvc8/ (12 files in 12 dirs): Initial check-in. |
14:32.49 | ``Erik | have ya figured out the full dependancy graph to figure out what classes you need to take first to keep your total stay short? |
14:33.15 | ``Erik | I had to load up on math early on, didn't do much cs the first year |
14:34.33 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/misc/win32-msvc8/ (btclsh/btclsh.vcproj bwish/bwish.vcproj): Increase optimization. |
14:34.33 | ``Erik | heh, by the time I finish an update and build, bob checks in smething else, so I have to do it again O.o :D |
14:34.54 | poolio | slow poke. |
14:34.59 | Defcon | :D |
14:35.29 | ``Erik | I'm doing a cvs -qz3 up -Pd at the top level dir, I'm sure if I'd cd into where he's making a mess, it'd go faster |
14:35.49 | ``Erik | and when he changes something like, y'know, raytrace.h ... |
14:37.32 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/misc/win32-msvc8/asc2g/asc2g.vcproj: Turn off debugging. |
14:39.26 | CIA-27 | BRL-CAD: 03bob1961 * 10brlcad/misc/win32-msvc8/brlcad/brlcad.sln: Add 12 more projects. |
14:46.28 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/NEWS: bob enabled/fixed the embedded framebuffer in the windows port and has added a slew of other missing utilities (too many to enumerate, but it's all of the ray-tracers). |
14:47.55 | ``Erik | bah |
14:48.28 | ``Erik | I finally get the winderz box back on the net, call hlepdesk to get my email passwd reset, and the entire site is having an email outage |
14:49.10 | ``Erik | accepting exchange for the calendar stuff was a mistake :/ |
14:49.23 | ``Erik | shoulda stuck with good old unix mail, it 'just works' |
15:30.41 | PrezKennedy | wish we could use some unix boxes here |
15:30.50 | PrezKennedy | for essential stuff |
15:35.36 | ``Erik | nice http://qdb.us/101461 |
15:36.37 | ``Erik | when I was at school, I got to watch 6 old aix boxes get replaced with "a couple" shiney new fast NT boxes |
15:37.00 | ``Erik | by the time the functionality was back to the same level for 'critical' services like email and web, there were over 300 nt servers |
15:40.43 | PrezKennedy | one for each person? |
15:40.44 | PrezKennedy | :-) |
15:41.09 | ``Erik | close to one for every ten people, I think, I dunno |
15:42.46 | ``Erik | (and the cs and math depts propped up their own servers, so they could stay useful... cs had a dual core fbsd beasty, math had a cheap desktop type running linux) |
15:55.42 | *** join/#brlcad prasad1 (n=psilva@static-70-108-244-218.res.east.verizon.net) |
16:29.42 | *** join/#brlcad minute (n=MinuteEl@silentflame/member/pdpc.base.minuteelectron) |
16:34.20 | *** join/#brlcad Elperion (n=Bary@p548759F3.dip.t-dialin.net) |
16:59.34 | Z80-Boy | ``Erik: there are serious problem with accuracy of what rt-edge produces |
16:59.41 | Z80-Boy | ``Erik: like it sometimes leaves out whole edges |
16:59.52 | Z80-Boy | ``Erik: is there any remedy to that? |
17:04.35 | ``Erik | um, make your thing bigger for raytracing? |
17:05.35 | *** join/#brlcad bpoole (n=bpoole@UNIX31.andrew.cmu.edu) |
17:11.54 | SkunkyFunky | ``Erik: but it leaves out edges which are not minuscule at all and the cylinders between which they are are also major cylinders in the model. |
17:13.02 | ``Erik | um, iirc, it looks for a neighbor hit distance and uses a fixed distance to see if it should call that an edge... and I think it might just look at horizontal neighbors :/ |
17:22.38 | SkunkyFunky | I |
17:22.46 | SkunkyFunky | I'll post the picture later the edge is 45 degrees |
17:31.32 | brlcad | SkunkyFunky: do you know the depth across the edge? |
17:31.42 | brlcad | it's an absolute size, iirc |
17:32.00 | brlcad | so if you're model is small, it won't see the edge |
17:32.14 | ``Erik | man that is freakin' ugly code :( |
17:32.19 | brlcad | there's not presently a run-time way to change that depth (would be a good mod to make) |
17:32.43 | brlcad | meh, didn't think it was all that horrible |
17:33.25 | brlcad | easy enough to understand regardless |
17:38.23 | SkunkyFunky | brlcad: 10mm or so |
17:38.28 | SkunkyFunky | and the thing is like 700mm big |
17:38.59 | brlcad | that's probably why |
17:39.08 | ``Erik | scale it up to rtedge |
17:39.10 | ``Erik | like |
17:39.10 | SkunkyFunky | How big is the hardcoded constant? |
17:39.17 | ``Erik | 10-20m big |
17:39.28 | SkunkyFunky | 10-20 meters? |
17:40.04 | ``Erik | yeah, that should fix it :D |
17:40.08 | SkunkyFunky | but that's a design relying on arbitrary constants, that cannot work reliably by principle |
17:40.18 | ``Erik | yes, it sucks, it needs to be fixed |
17:41.14 | SkunkyFunky | can it be done without the constant by principle at all? |
17:41.33 | SkunkyFunky | does anyone have jdk-1_5_0_10-linux-i586.bin? |
17:41.41 | brlcad | SkunkyFunky: it's not a hardcoded constant |
17:41.43 | SkunkyFunky | LFS needs it but the Sun downloads don't have it anymore/ |
17:41.47 | brlcad | it's based on the cell size |
17:41.49 | SkunkyFunky | brlcad: where can it be set? |
17:41.53 | SkunkyFunky | brlcad: what's a cell size? |
17:41.58 | brlcad | in the code, look for this: |
17:41.59 | brlcad | <PROTECTED> |
17:42.04 | brlcad | <PROTECTED> |
17:42.09 | brlcad | <PROTECTED> |
17:42.17 | brlcad | <PROTECTED> |
17:42.19 | brlcad | <PROTECTED> |
17:42.38 | SkunkyFunky | where does 87 come from? An universal extraterrestrial constant? |
17:43.28 | brlcad | would have to look at lgt |
17:43.50 | brlcad | probably just derived from experimental sampling of vehicles |
17:44.06 | brlcad | value is of course, 19.08 |
17:44.09 | SkunkyFunky | My Ronja is not a vehicle that's the problem |
17:44.13 | brlcad | yep |
17:44.16 | brlcad | a long known issue |
17:44.23 | SkunkyFunky | How many wheels do I have to implement to become a vehicle? |
17:44.28 | brlcad | like I said, it'd be a great mod to make |
17:44.52 | SkunkyFunky | I could make a field Ronja with 4 balooney wheels |
17:45.10 | SkunkyFunky | or with caterpillar tracks for better traction in swampy terrains |
17:45.15 | brlcad | wasn't just vehicles, probably just a random sampling of models |
17:45.23 | brlcad | but then I'm just entirely guessing |
17:45.37 | brlcad | rtedge guy took it from lgt, lgt does back 15+ years |
17:45.55 | SkunkyFunky | and back then they tossed a coin |
17:46.15 | SkunkyFunky | or a cow or whatever was used for payment at that times ;-) |
17:47.00 | brlcad | flipping a coin and get 87, don't think so |
17:47.38 | SkunkyFunky | I just got an idea of acoustic burglar alarm based on speaker-microphone feedback |
17:48.04 | SkunkyFunky | Put a microphone and speaker into some space and when a person passes they create a reflection which reflects with >1 loop gain and oscillation is created |
17:48.17 | SkunkyFunky | brlcad: maybe they flipped a 100 dollar bill and got 87? |
17:48.26 | SkunkyFunky | from the devaluation? ;-) |
17:49.14 | SkunkyFunky | or 2 speakers transmitting in opposite phase and a microphone just between them |
17:49.37 | SkunkyFunky | with high gain. In clear state it zeroes out almost perfectly and when someone breaks the acoustic field it starts a shrill |
17:49.39 | brlcad | commit message from gary moss (back in '88) was simply "Added distance discriminant map to hiddenline drawing model." |
17:50.00 | SkunkyFunky | brlcad: isn't this constant an inevitable part of the algorithm? |
17:50.13 | SkunkyFunky | brlcad: or is it a way to identify an edge without any constant? |
17:50.21 | SkunkyFunky | I mean of course dynamically calculated constant as you said |
17:50.42 | brlcad | either way, it's been a long-known limitation that it's cellsize dependent -- what it means, though, is that you may be able to generate a bigger image to get the edge (add an edge thickness) |
17:51.16 | brlcad | SkunkyFunky: edges are determined (in rtedge) via multiple checks, depth simply being the main driver for most models |
17:51.18 | SkunkyFunky | brlcad: bigger image you mean more pixels by more pixels? |
17:51.26 | brlcad | sharp changes in curvature will do it too |
17:51.31 | brlcad | yes |
17:51.53 | SkunkyFunky | brlcad: I cannot find the edge thickness in brlman rtedge |
17:52.16 | SkunkyFunky | -c Set special rtedge configuration variables |
17:52.20 | SkunkyFunky | Is it one of them? |
17:52.56 | brlcad | the fix is to just make it a settable depth since there's not really a good way to determine the edges desired for arbitrary models without brep conversion |
17:53.01 | CIA-27 | BRL-CAD: 03erikgreenwald * 10brlcad/src/rt/viewedge.c: minor clean-up |
17:53.05 | brlcad | yeah, it's a -c var |
17:53.59 | brlcad | i think |
17:54.14 | SkunkyFunky | can't find it |
17:55.32 | brlcad | ah, my mistake, no such option |
17:55.48 | SkunkyFunky | :( |
17:55.51 | SkunkyFunky | That would be handy |
17:56.05 | brlcad | you can still generate the larger image and it should give the edge if sufficiently big |
17:56.20 | brlcad | assuming you don't exceed the max image dimensions of course |
17:56.26 | SkunkyFunky | yes but how do I make the edge thicker before downsampling the image to overlay with the raytraced model? |
17:56.31 | SkunkyFunky | which is? |
17:56.36 | SkunkyFunky | (another hardcoded constat?) |
17:56.56 | SkunkyFunky | is there some pixspread or pixerode program? |
17:57.00 | SkunkyFunky | or pixgrow |
17:57.08 | SkunkyFunky | pixshrink |
17:58.23 | brlcad | the max "should" be uint x uint (so about 2billion x 2billion) but there are various temp buffers throughout that haven't been all weeded out yet, so practically it's probably more like 64k x 64k |
17:59.07 | brlcad | either way, it's more than photoshop could handle before CS came out :) |
17:59.09 | SkunkyFunky | that's enough |
17:59.19 | SkunkyFunky | Counterstrike? |
17:59.25 | brlcad | Photoshop CS |
17:59.33 | SkunkyFunky | Czechoslovak edition? |
17:59.43 | brlcad | beats me, google it up |
18:00.22 | brlcad | for your purpose, you really could just use the configurable depth |
18:00.34 | brlcad | there ya go |
18:01.26 | SkunkyFunky | brlcad: but the the edges are jagged anyway |
18:01.38 | SkunkyFunky | often looks good but often looks ugly too |
18:01.44 | CIA-27 | BRL-CAD: 03erikgreenwald * 10brlcad/src/rt/viewedge.c: allow users to set arbitrary distance threshholds |
18:01.48 | SkunkyFunky | is the rtedge substantially faster than rt? |
18:01.55 | SkunkyFunky | ``Erik: Thanks :) |
18:02.13 | SkunkyFunky | O.O |
18:03.10 | brlcad | SkunkyFunky: yes, rtedge does considerably less work than rt |
18:03.20 | brlcad | so it's faster in what it does |
18:05.18 | SkunkyFunky | brlcad: did you find out why that model was 291x slower than the other ones? |
18:06.12 | ``Erik | think that deserves a NEWS line? |
18:06.13 | brlcad | the reason was exactly what john mentioned |
18:06.17 | brlcad | ``Erik: definitely does |
18:06.48 | brlcad | every user-visible non-dev change |
18:07.01 | brlcad | don't forget the manpage ;) |
18:07.03 | CIA-27 | BRL-CAD: 03erikgreenwald * 10brlcad/src/rt/rtedge.1: mention the new max_dist option |
18:07.19 | ``Erik | huh? |
18:07.20 | ``Erik | the what? |
18:07.48 | brlcad | :) |
18:08.03 | ``Erik | <-- had already hit enter on the commit when he asked if it was worth NEWS... :) |
18:08.07 | CIA-27 | BRL-CAD: 03erikgreenwald * 10brlcad/NEWS: mention max_dist option for rtedge |
18:09.47 | ``Erik | some, um, interesting little details... it doesn't buffer anything up, so the edge 'detection' only uses two pixels, the one immediately below and the one to the left O.o |
18:10.32 | brlcad | yeah, ron talked about that way back when |
18:11.00 | brlcad | it was enough just to get it working at the time, he had all sorts of plans for improving it.. |
18:11.09 | ``Erik | heh |
18:41.33 | CIA-27 | libirc: 03JeffM2501 * r313 10/trunk/libirc/examples/stupidBot/src/stupidBot.cpp: |
18:41.33 | CIA-27 | libirc: save off the echo settings on a flush |
18:42.01 | CIA-27 | libirc: add command to set echos in realtime |
19:31.25 | *** join/#brlcad yukonbob (n=yukonbob@198.235.198.234) |
19:31.46 | yukonbob | hello, whirled |
20:18.50 | *** join/#brlcad iday (n=jlowens@bz.bzflag.bz) |
20:23.17 | *** join/#brlcad Z80-Boy (n=clock@77-56-74-13.dclient.hispeed.ch) |
20:24.40 | *** join/#brlcad cad47 (n=422090d9@bz.bzflag.bz) |
20:38.22 | Z80-Boy | brlcad: actually rt takes 1.6 sec and rtedge 3.18 sec CPU time |
20:38.52 | Z80-Boy | I have noticed there seems to be some kind of preparation phase before the raytracing itself. And this takes significant amount of time. |
20:39.10 | Z80-Boy | Can this time be saved if I have the same geometry and output just different azimuths? |
20:46.18 | Z80-Boy | clock@kestrel:~$ anim_script -a 10 |
20:46.18 | Z80-Boy | Segmentation fault (core dumped) |
20:46.19 | Z80-Boy | Ouch |
21:59.52 | *** join/#brlcad elite01 (n=elite01@dslc-082-082-065-178.pools.arcor-ip.net) [NETSPLIT VICTIM] |
21:59.53 | *** join/#brlcad SWPadnos (n=Me@emc/developer/SWPadnos) [NETSPLIT VICTIM] |
22:00.58 | *** join/#brlcad brlcad (n=sean@pdpc/supporter/silver/brlcad) [NETSPLIT VICTIM] |
22:00.58 | *** join/#brlcad PrezKennedy (i=Matt@74.86.45.130) [NETSPLIT VICTIM] |
22:00.58 | *** join/#brlcad ``Erik (i=erik@c-68-54-174-162.hsd1.md.comcast.net) [NETSPLIT VICTIM] |
22:00.58 | *** join/#brlcad Maloeran (n=maloeran@glvortex.net) [NETSPLIT VICTIM] |
22:00.58 | *** join/#brlcad CIA-27 (n=CIA@208.69.182.149) [NETSPLIT VICTIM] |
22:00.58 | *** mode/#brlcad [+o brlcad] by irc.freenode.net |
22:01.03 | *** join/#brlcad starseeker (n=CY@ip72-218-18-244.hr.hr.cox.net) [NETSPLIT VICTIM] |
22:01.45 | *** join/#brlcad starseeker (n=CY@ip72-218-18-244.hr.hr.cox.net) |
22:01.46 | starseeker | wee |
22:08.44 | ``Erik | aa heh |
22:08.51 | ``Erik | rfc1459, yo :D |
22:11.21 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/configure.ac: check for fchmod() |
22:11.40 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/include/config_win.h: don't have fchmod or _fchmod apparently |
22:14.49 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/src/other/iwidgets/pkgIndex.tcl: according to the sources, this is still 4.0.1 |
22:16.47 | CIA-27 | BRL-CAD: 03brlcad * 10brlcad/src/libbu/fchmod.c: rework bu_fchmod() to use HAVE_FCHMOD feature, falling back to chmod if it's not available. add missing footer. convert to unix line endings. |
22:18.50 | *** join/#brlcad docelic (n=docelic@77.237.110.131) |
22:50.00 | CIA-27 | BRL-CAD: 03erikgreenwald * 10brlcad/src/bwish/ (cadAppInit.c winMain.c): dos->unix line terminators |
22:50.18 | ``Erik | <-- not gonna dork with the 3 in libz |
23:03.57 | Z80-Boy | Some outlined drawings |
23:03.58 | Z80-Boy | http://ronja.twibright.com/3d/par_welded_cut_0.png |
23:04.30 | Z80-Boy | http://ronja.twibright.com/3d/headcut_0.png |
23:16.14 | brlcad | Z80-Boy: I figured you were doing that :) |
23:16.15 | brlcad | nice |
23:16.25 | brlcad | have you played with rtwizard? |
23:21.28 | bpoole | alloo |
23:22.17 | Z80-Boy | brlcad: no |
23:22.46 | Z80-Boy | Also this http://ronja.twibright.com/3d/railing_0.png |
23:23.16 | Z80-Boy | and this http://ronja.twibright.com/3d/plazmatron_0.png |
23:23.35 | brlcad | alloo bpoole |
23:24.13 | Z80-Boy | and this http://ronja.twibright.com/3d/ronja_new_0.png |
23:24.42 | brlcad | yeah, the main reason it's drawing the edges that it's drawing is because the normal vector for the neighbors is drasticly different |
23:25.13 | brlcad | the ones it's not drawing have the same orientation |
23:25.36 | Z80-Boy | Here you can see where the edge is left out on the cylinders between brown and grey http://ronja.twibright.com/3d/ronja_0.png |
23:26.27 | brlcad | hep, the curvature matchs |
23:26.31 | Z80-Boy | On this one you can see it's making up dots on the rim http://ronja.twibright.com/3d/rear_cap_0.png |
23:26.44 | brlcad | the more you go up that curve, the more they diverge |
23:27.32 | Z80-Boy | on this one you can see some kind of double-dotted line it made up in a place where is no edge: http://ronja.twibright.com/3d/mast_0.png |
23:27.52 | bpoole | brlcad: so, thanksgiving I think I'm going to re-read my code and try to maybe start working on beset again, for real this time :) |
23:27.55 | Z80-Boy | If it considers a transition from straight plane into a curved one an edge then it should print a solid edge and not double dotted one |
23:28.36 | Z80-Boy | The pattern is actually .. . .. .. . .. .. . .. .. . or something like that interesting |
23:28.53 | brlcad | it's because of the change in the normal |
23:29.01 | Z80-Boy | And also made up couple of randomly placed dots around the object |
23:29.03 | brlcad | floating point fuzz -- it's right on the edge |
23:29.25 | brlcad | tweaking the curvature angle would make that line solid or disappear |
23:29.36 | brlcad | (not in the model, in rtedge) |
23:29.45 | Z80-Boy | there is no angle the rounded surface is tangential to the straight one |
23:30.05 | brlcad | there is a normal angle |
23:30.14 | brlcad | for every pixel that hits something |
23:30.20 | Z80-Boy | and what are the two black dots just above the top edge on the white U channel? |
23:30.23 | brlcad | it uses that angle to determine hit |
23:30.35 | Z80-Boy | There is nothing at that place that could be misinterpreted it's just a solid rpp |
23:32.21 | brlcad | not obvious for the points on the top, not clear, could be the edge of some negative primitive that is getting sorted in front |
23:32.34 | brlcad | causing a perturbed normal for just those few pixels |
23:32.57 | brlcad | related to the issue yukonbob was seeing in his render |
23:33.29 | brlcad | the other "spots" though are on a a curvature end/start point where there is a change |
23:33.41 | Z80-Boy | But there is no doubt that if you have a straight surface where there are no edges or seams between primitives that there is no edge! |
23:34.30 | brlcad | i'm not disagreeing with you on the top ones |
23:34.45 | brlcad | the ones, however, where curvature starts, however, is quite debateable |
23:34.51 | Z80-Boy | And the little yellow edge right of the blue pipe is partly missing, although the shading on both sides is visibly different |
23:34.59 | Z80-Boy | Is it because normals are too similar? |
23:35.00 | brlcad | most cad systems *will* show you bot the beginning and end of that inside blend |
23:35.30 | Z80-Boy | Well they should either show nothing or a solid edge and not a broken one |
23:35.44 | brlcad | sure |
23:35.49 | Z80-Boy | cause the whole interface is translation symmetric so the output should be also translation symmetric |
23:35.58 | brlcad | it's just numerically working out that way due to floating point fuzz |
23:36.25 | brlcad | there's very little you can do about it when you're on the boundary of the curvature value, other than change the value |
23:36.37 | Z80-Boy | I thought all those minimum-angle minimum-distance constatnts are there to suppress the floating point fuzz |
23:36.46 | Z80-Boy | BTW floating point behaves in an astonishing way. |
23:37.04 | brlcad | it's not there to suppress floating point, it's there to detect the edge |
23:37.17 | Z80-Boy | For example float a; <something> printf("%G\n",a); printf("%G\n",a); once gave me two different results which differed in the last digit |
23:37.20 | brlcad | we're dealing with implicit prims, so there is no actual edge other than the one sampled |
23:37.40 | brlcad | that's not astonishing, it's the way it is |
23:37.48 | Z80-Boy | Yes, the value of the floating point variable can change without any write into the variable! |
23:38.20 | brlcad | you would also then be astonished how much code is actually dedicated to trying to manage floating point fuzz in all the computations... |
23:38.40 | Z80-Boy | I am not surprised I once tried to write a doom-like game engine ;-) |
23:38.49 | Z80-Boy | That's where I stroke that illogical behaviour |
23:38.59 | brlcad | yes, it can, particularly if it's not a representable value and might depend upon which registers it's loaded into or how the floating point unit operates |
23:39.50 | Z80-Boy | it was a normal number |
23:39.53 | Z80-Boy | say like 1.38 or so |
23:40.05 | brlcad | that doesn't mean it's representable |
23:41.21 | brlcad | 1000.2 is not a representable number, for example, for a float |
23:41.30 | brlcad | yet it's perfectly "normal" |
23:42.29 | Z80-Boy | If it were calculated in fixed point then at least a==a would be valid |
23:42.41 | Z80-Boy | and not that it would print two different numbers without a write into the variable. |
23:43.07 | Z80-Boy | Floating point behaves a bit like analogue compuiter |
23:43.57 | brlcad | even .1 isn't representable |
23:45.33 | Z80-Boy | .1 .2 .3 .4 .6 .7 .8 .9 |
23:46.57 | Z80-Boy | Real numbers are just a school illusion |
23:47.01 | Z80-Boy | As well as integers. |
23:48.06 | Z80-Boy | Kids are taught to think about the world in terms of something that cannot exist by principle |
23:49.10 | brlcad | it exists within a tolerance ;) |
23:51.23 | Z80-Boy | Here the results are not very good either http://ronja.twibright.com/3d/hardware_1.png |