| 00:11.23 | *** join/#brlcad nmz787_i1 (nmccorkx@nat/intel/x-lsyzidbvmqpccxdl) |
| 00:20.33 | starseeker | I wonder if that openscad STL import is making use of this capability: http://docs.pointclouds.org/trunk/group__sample__consensus.html |
| 00:35.26 | *** join/#brlcad joythewizard (43aff2ac@gateway/web/freenode/ip.67.175.242.172) |
| 00:53.06 | *** join/#brlcad YatharthROCK (~YatharthR@unaffiliated/yatharthrock) |
| 01:09.06 | Notify | 02GCI:zhang_yizhuo * 4992155881832448 : Task Claimed - I would like to work on this task. |
| 01:10.32 | Notify | 02GCI:joy818200 * 6384883232407552 : Task Claimed - I would like to work on this task. |
| 01:13.52 | Notify | 02GCI:zhang_yizhuo * 4992155881832448 : None - Hi, I have sent my name & postcode to contest@brlcad.org. The first four letters of my email are zhan. Thank you. |
| 01:26.34 | Notify | 02GCI:goobypls * 5269456200663040 : Task Claimed - I would like to work on this task. |
| 02:02.03 | brlcad | andromeda-galaxy: more sample pattern work would be higher priority over both those |
| 02:02.49 | brlcad | we need a function that sets up a grid of rays given a view just like you did for cone |
| 02:03.01 | brlcad | orthogonal and projective versions |
| 02:03.25 | brlcad | (which are prism and rectangle shapes) |
| 02:03.48 | brlcad | along with updating our ray tracers to use that logic instead of the custom logic they use now |
| 02:21.56 | Notify | 02GCI:devikakumar * 4639899508539392 : Task Claimed - I would like to work on this task. |
| 02:32.18 | Notify | 02GCI:adityagulati * 5581079297654784 : Ready for review - The work on this task is ready to be reviewed. |
| 02:33.58 | andromeda-galaxy | brlcad: makes sense.. did you add any more tasks for the sample patterns? I couldn't find any |
| 02:52.34 | Notify | 02GCI:brlcad * 4992155881832448 : Task Assigned - This task has been assigned to Yizhuo. You have 100 hours to complete this task, good luck! |
| 02:52.35 | Notify | 02GCI:brlcad * 5269456200663040 : Task Assigned - This task has been assigned to goobypls. You have 100 hours to complete this task, good luck! |
| 02:52.49 | Notify | 02GCI:brlcad * 6384883232407552 : Task Assigned - This task has been assigned to Joy Chen. You have 100 hours to complete this task, good luck! |
| 02:56.34 | nmz787 | kanzure: brlcad this is interesting, since I am interested in making sine wave type things http://math.stackexchange.com/questions/116369/can-a-rational-b%C3%A9zier-curve-take-exactly-the-same-shape-as-a-part-of-the-sine-fu |
| 02:57.08 | nmz787 | "In a Bézier curve, x and y are polynomials in the parameter t. Note that you can't just have "a part of the sine function": if y(t)=sin(x(t)) for t in some interval, since both sides of that equation are analytic functions on the complex plane the equation would be true for all complex numbers t. Since y(t) is a polynomial, for any given value of y (unless y is constant) there are only finitely many t and thus finitely many x. But this is not ... |
| 02:57.14 | nmz787 | ... the case for the sine function: sin(nÏ)=0 for all integers n. So the sine curve can't be given exactly by a Bézier curve of any degree." |
| 02:58.01 | nmz787 | does the same logic apply to nurbs? |
| 03:01.38 | andromeda-galaxy | brlcad: I want to claim a raybundle task or one of the project tasks, of the tasks that are currently up, which is highest priority? |
| 03:04.19 | andromeda-galaxy | brlcad: (I can't find any bundle tasks in the current list) |
| 03:06.39 | andromeda-galaxy | <PROTECTED> |
| 03:14.21 | kanzure | nmz787: with nurbs i highly recommend not using the underlying nurbs implementation and instead using geometric primitives designed from nurbs (like csg parts or otherwise) |
| 03:21.11 | nmz787 | kanzure: someone earlier was saying CSG was some sort of tree structure, didn't sound like it was nurbs |
| 03:24.24 | kanzure | yeah that's bullshit, you can do csg out of anything |
| 03:24.36 | kanzure | trees are very common structures in software |
| 03:27.51 | nmz787 | so does the primitives in mged default to use nurbs, then fallback to something else? |
| 03:28.00 | nmz787 | or if not mged, which brl tool? |
| 03:28.11 | nmz787 | I've so far only seen stuff relating to mged |
| 03:28.20 | nmz787 | (examples of models) |
| 03:39.11 | andromeda-galaxy | brlcad, maths22: I'm curious about the commit-testing web application project, but am curious as to what web server brlcad.org is using, it seems like making live updating work would require having some kind of websockets implementation on the backend... also, having some method sandboxing scripts seems like a good idea... |
| 03:44.32 | kanzure | nmz787: that's a good question, my understanding is that the default is not nurbs in brlcad, and there is a way to access nurbs somehow... check src/libbrep/ |
| 03:44.51 | kanzure | like /src/libbrep/test_curve_intersect |
| 03:44.52 | brlcad | andromeda-galaxy: there are now two additional ray sampling tasks, an updated vis task, and two integration tasks posted (along with 20+ other new tasks) |
| 03:45.33 | brlcad | search on "sampl" and "vis" |
| 03:47.36 | kanzure | nmz787: this person here is a good person to ask these questions to |
| 03:47.41 | kanzure | nmz787: brlcad might even know things about brlcad |
| 03:48.30 | andromeda-galaxy | brlcad: I don't see them, searching on 'sampl' |
| 03:48.39 | andromeda-galaxy | do you have any links? |
| 03:48.59 | andromeda-galaxy | and 'vis' only shows the compile with msvc++ task |
| 03:49.52 | brlcad | andromeda-galaxy: the argument is pretty much correct, you can only approximate something smooth like a sine curve with a polynomial curve -- you'd need an infinite sum of curves iirc, taylor series expansion |
| 03:50.07 | brlcad | that said, for all practical purposes, it won't matter |
| 03:50.33 | andromeda-galaxy | brlcad: that's true... it's just phrased in an interesting manner --- that argument makes it sound like a portion of a curve should be doable |
| 03:50.33 | brlcad | you'll be "tight-enough" probably with a sixth order spline |
| 03:52.10 | brlcad | nmz787: I suggest reading brlcad.org/wiki/Documentation, particularly the mged tutorials (which can be completed in just a few hours) -- that will give you a great overview of some really basic concepts |
| 03:53.12 | brlcad | andromeda-galaxy: we have full control over our server |
| 03:54.21 | brlcad | nmz787: kanzure is correct -- we don't default to nurbs, but almost any primitive will convert flawlessly to nurbs format. we're working on nurbs boolean operations now (some work, some don't). |
| 03:54.23 | andromeda-galaxy | brlcad: makes sense... I was just thinking that keeping dependencies at a minimum + having nice live updating using websockets/long polling might be hard, since for those you often essentially need an application web server |
| 03:55.30 | andromeda-galaxy | brlcad: okay, I see them now |
| 03:55.32 | brlcad | eh, you put "sampl" in the title field? I see four with that |
| 03:55.34 | andromeda-galaxy | they weren't showing up earlier for some reason |
| 03:55.38 | brlcad | okay |
| 03:55.41 | andromeda-galaxy | sorry |
| 03:56.01 | brlcad | you have to refresh, list doesn't do it for you even though it seems to update dynamically (it doesn't) |
| 03:56.29 | brlcad | also one vis task #2, this time using geometry or plot3 ;) |
| 03:56.51 | andromeda-galaxy | brlcad: makes sense |
| 03:57.09 | andromeda-galaxy | should I start working on another implementation, or a visualization? |
| 03:57.21 | brlcad | the two new patterns are quite simple -- the work is really in the integration and making sure args are appropriate for hooking them into rt* tools |
| 03:57.28 | brlcad | vis is last |
| 03:57.31 | andromeda-galaxy | sure |
| 03:57.54 | andromeda-galaxy | so implement, then utilize, then vis? |
| 03:58.24 | andromeda-galaxy | the vis one will be somewhat interesting, I don't have too much experience with the debugging infrastructure yet |
| 03:58.55 | andromeda-galaxy | by the way, rt seems to have *lots* of global variables... is there a plan to eventually refactor the whole tool? |
| 04:00.28 | andromeda-galaxy | brlcad: so for sampling a frustum, |
| 04:00.31 | brlcad | implement, vis, utilize |
| 04:00.39 | andromeda-galaxy | ah, okay |
| 04:00.49 | andromeda-galaxy | should there be a and b vectors and then a and b divergence angles? |
| 04:01.22 | andromeda-galaxy | a lot like the cone from earlier except a) a rectangular grid and b) having separate values for a-direction and b-direction |
| 04:01.25 | andromeda-galaxy | ? |
| 04:01.28 | brlcad | http://upload.wikimedia.org/wikipedia/commons/thumb/8/83/Ray_trace_diagram.svg/2000px-Ray_trace_diagram.svg.png |
| 04:02.22 | andromeda-galaxy | hmm? I'm just checking on the geometry of the frustum that you meant... |
| 04:03.02 | brlcad | we use almost the same terms as these guys (albeit in a different default "up"): http://wiki.povray.org/content/Reference:Camera |
| 04:03.33 | nmz787 | brlcad: thanks, I'll check it out this weekend. Is there by chance an API style documentation online somewhere? I saw some mention of doxygen, but didn't see what looked like doxy-generated HTML on the web site. |
| 04:04.05 | nmz787 | (also as a comment, I was confused that some of the stuff under the Wiki tab, was not linked to in the Docs section of the site) |
| 04:04.08 | andromeda-galaxy | brlcad: is the up vector user-specifiable? |
| 04:04.20 | brlcad | andromeda-galaxy: it sounds like you have the right idea -- the tough part is to make sure you start with the right view information (or could wrap your function with a view function) |
| 04:04.33 | brlcad | the entire view is user-specified |
| 04:04.47 | andromeda-galaxy | right... so I was saying above, should it support having different diveregence angles |
| 04:04.52 | andromeda-galaxy | in the 'up' and 'sideways' directions? |
| 04:05.26 | brlcad | it needs to support different divergence angles, yes |
| 04:05.37 | brlcad | maybe look at what the rt* apps do first |
| 04:05.46 | andromeda-galaxy | good idea, one second... |
| 04:05.53 | brlcad | their code is pretty much exactly what you want because the very next step is to replace what they do with your function |
| 04:06.05 | andromeda-galaxy | after that, if you're still around, I just want to quickly check the api for it with you |
| 04:06.37 | brlcad | the only problem being that the ray dispatch logic is mixed in with a whole bunch of other stuff and lots of globals are involved, and it's messy old code with dragons in there ;) |
| 04:07.22 | brlcad | I'm out for the night, but feel free to e-mail the list and I can respond when I can (or maybe someone else will if your question is simple enough and/or your description clear enough |
| 04:07.59 | andromeda-galaxy | brlcad: sure, but will you be around for the next couple of minutes? I can probably come up with a first version in the next few minutes after reading do_frame( |
| 04:08.02 | andromeda-galaxy | ) |
| 04:09.57 | brlcad | in all honestly, you're good but if it only takes you a couple minutes, you've almost certainly not put enough time into reading the code ... it's pretty complicated to set up a fully parameterized view |
| 04:10.35 | andromeda-galaxy | true.. I'm looking through it now, anad it's more complicated than I thought at first |
| 04:10.37 | brlcad | there's the dimensions, divergence, aspect ratio, ortho or persp angle, az/el/quat orientation , ... |
| 04:10.44 | andromeda-galaxy | so manynnested worker threads... |
| 04:10.58 | andromeda-galaxy | it seems like a sane api for the function would be something like (without having finished rt) |
| 04:11.25 | andromeda-galaxy | bu_gen_frustum(vect_t a, vect_t b, fastf_t a_divergence, fastf_t b_divergence, fastf_t a_nurays, fastf_t b_numrays) |
| 04:11.33 | brlcad | will almost certainly help to spend about 10 minutes and read the rt manual page too, to read all the options that pertain to the view |
| 04:11.40 | andromeda-galaxy | oh, and center of course |
| 04:12.09 | nmz787 | no API/doxygen stuff online then? |
| 04:12.24 | brlcad | andromeda-galaxy: that would be my guess too and there'd probably be another function to convert user-specifications into that form |
| 04:12.49 | andromeda-galaxy | brlcad: right, or maybe the function calling it from main rt could marshall up all the globals and convert them nicely |
| 04:13.03 | brlcad | nmz787: we do somewhere, but our APIs (plural, we have about 2 dozen) are quite large -- best to start with a specific task in mind :) |
| 04:13.22 | nmz787 | brlcad: I like to browse to know what kind of ideas are realistic |
| 04:14.19 | nmz787 | brlcad: the documentation page doesn't seem to mention 24 programs, I see mged, oed, nirt |
| 04:15.02 | andromeda-galaxy | brlcad: would it be possible to ideally do away with the worker()/do_run() stuff and replace it with shooting a bundle with rt_shootrays()? |
| 04:15.05 | brlcad | nmz787: our basic utility library along has something like 300 functions across about three dozen concepts, a couple other libs have 1-2 thousand public functions |
| 04:15.20 | brlcad | nmz787: we don't like to drown people in information (we easily can, as can most CAD systems) |
| 04:15.27 | brlcad | we have over 400 binaries |
| 04:15.28 | andromeda-galaxy | (that is, I think that that is the ultimate objective, but want to theck) |
| 04:15.52 | brlcad | granted, only about 10% of those are used 90% of the time ;) |
| 04:16.30 | nmz787 | brlcad: so this is openly-obscure source? :) |
| 04:16.31 | brlcad | nmz787: if you want to browse, I suggest looking at http://en.flossmanuals.net/contributors-guide-to-brl-cad/ |
| 04:16.58 | kanzure | nmz787: there are books/documentation somewhere |
| 04:16.59 | *** join/#brlcad YashM (~YashM@117.198.4.129) |
| 04:18.08 | nmz787 | I just don't understand why people would work on doxygen stuff, if no one uploads it and adds a link to the website :/ |
| 04:18.29 | brlcad | andromeda-galaxy: that is certainly ideally possible but ... it's complicated as worker is also the dispatcher and there are many different types of dispatch patterns |
| 04:18.47 | andromeda-galaxy | brlcad: true... |
| 04:19.01 | nmz787 | huh, I have a lot more time in using PING than BRLCAD, hopefully that will change soon :) |
| 04:19.01 | brlcad | andromeda-galaxy: not even sure yet how they will come into play here (e.g., shooting line at a time, stamp at a time, recursive refinement, etc) |
| 04:19.03 | andromeda-galaxy | so the alternative is to leave it using worker as it is now, but replace the ray generation logic with the bundle logic |
| 04:19.15 | andromeda-galaxy | ? |
| 04:19.15 | brlcad | reading the manual page really should help shed some light |
| 04:19.39 | andromeda-galaxy | true, last time I looked at it, it didn't seem to describe all the view parameters... |
| 04:19.43 | brlcad | nmz787: we have it integrated into our code so if you've downloaded it, there's a build target for doxygen iirc |
| 04:20.04 | kanzure | nmz787: i think you want the mged quick reference pdf |
| 04:20.11 | nmz787 | brlcad: how do I submit a bug report or feature request to have someone upload the HTML? |
| 04:20.14 | brlcad | nmz787: here's an old output that was uploaded http://brlcad.sourceforge.net/doxygen/ |
| 04:20.59 | nmz787 | kanzure: not /just/ that though... there are 300 some binaries or something |
| 04:21.04 | kanzure | many more |
| 04:21.15 | brlcad | nmz787: brl-cad was started by the same guy that wrote ping ;) |
| 04:21.33 | nmz787 | brlcad: that was my point, I just learned that from kanzures link |
| 04:21.39 | brlcad | ahh, gotcha |
| 04:21.51 | kanzure | what link |
| 04:21.53 | brlcad | kanzure's link? which was that? |
| 04:22.11 | brlcad | blames it on kanzure |
| 04:22.24 | nmz787 | http://en.flossmanuals.net/contributors-guide-to-brl-cad/ |
| 04:22.33 | kanzure | hehehe |
| 04:22.33 | brlcad | i suggested that |
| 04:22.37 | nmz787 | err |
| 04:22.39 | nmz787 | oh yeah |
| 04:22.40 | nmz787 | that one |
| 04:22.42 | nmz787 | heh |
| 04:22.43 | nmz787 | sorry |
| 04:22.47 | kanzure | it's easy to mistake us |
| 04:22.49 | nmz787 | i am burned out from the day at work |
| 04:22.56 | brlcad | he's the handsome one |
| 04:23.04 | nmz787 | hah |
| 04:23.08 | kanzure | and he's the one trained in deadly weapons technology and military awesomeness |
| 04:23.33 | kanzure | goes back to writing tests |
| 04:23.45 | brlcad | goes to saw some wood |
| 04:24.17 | nmz787 | goes to do chemistry homework |
| 04:26.29 | andromeda-galaxy | goes to work on understanding rt's global-ridden code |
| 04:29.29 | brlcad | implemented in a time long long ago when globals in *applications* were not frowned upon in the least |
| 04:30.14 | brlcad | we're on a mission to eliminate them all slowly, though |
| 04:35.04 | andromeda-galaxy | ah... well, good to know that they'll (hopefully) go at some point.. |
| 04:35.34 | andromeda-galaxy | converting rt to use bundled rays is going to be .... interesting |
| 04:35.52 | andromeda-galaxy | as it is, it appears that it defers the computation of the ray boundaries into do_pixel() |
| 04:35.57 | andromeda-galaxy | which is also important for hypersampling |
| 04:36.59 | andromeda-galaxy | brlcad: except that parts of it are set up in setup_grid() |
| 04:37.40 | andromeda-galaxy | so it would have to do something like compute the bundle in do_frame() and the have do_pixel() manage to get the right pixel out of it instead of computing its own... |
| 06:54.15 | *** join/#brlcad adityagulati (dce32db2@gateway/web/cgi-irc/kiwiirc.com/ip.220.227.45.178) |
| 07:51.29 | *** join/#brlcad infobot (ibot@rikers.org) |
| 07:51.29 | *** topic/#brlcad is Topic for #brlcad: BRL-CAD || http://brlcad.org || logs: http://ibot.rikers.org/%23brlcad/ || Welcome GCI 2014 participants! If you're participating, be patient when asking questions... we're very busy and sometimes have packed schedules. We will respond if you are still on IRC: learn screen+irssi. |
| 07:52.39 | *** join/#brlcad YashM (~YashM@59.88.24.248) |
| 08:08.27 | *** join/#brlcad raptor (~raptor@194.28.75.221) |
| 08:10.53 | *** join/#brlcad albertcoder (~quassel@202.164.53.117) |
| 09:06.25 | raptor | Help stared at the screen . I can not continue to draw. http://keep4u.ru/full/6fe928c75b85e577419f35cde5a37912.html |
| 09:09.12 | raptor | brlcad, Help stared at the screen . I can not continue to draw. http://keep4u.ru/full/6fe928c75b85e577419f35cde5a37912.html |
| 09:33.50 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 09:55.28 | *** join/#brlcad luca79 (~luca@host109-111-dynamic.4-87-r.retail.telecomitalia.it) |
| 09:57.51 | *** join/#brlcad arno (~luca@host188-230-dynamic.5-87-r.retail.telecomitalia.it) |
| 10:00.55 | *** join/#brlcad luca79 (~luca@host188-230-dynamic.5-87-r.retail.telecomitalia.it) |
| 10:36.50 | *** join/#brlcad luca79 (~luca@host188-230-dynamic.5-87-r.retail.telecomitalia.it) |
| 11:04.59 | *** join/#brlcad merzo (~merzo@121-34-132-95.pool.ukrtel.net) |
| 11:10.08 | *** join/#brlcad luca79 (~luca@host188-230-dynamic.5-87-r.retail.telecomitalia.it) |
| 11:35.32 | *** join/#brlcad raptor_ (c21c4bdd@gateway/web/freenode/ip.194.28.75.221) |
| 11:46.05 | raptor | How to set the default perspective ? |
| 11:46.26 | raptor | I accidentally knocked |
| 11:52.18 | *** join/#brlcad YashM (~YashM@117.223.174.28) |
| 12:19.39 | raptor | <PROTECTED> |
| 12:32.11 | *** join/#brlcad jasvir (~jasvir@122.173.150.75) |
| 12:48.28 | ignacio | Good morning :) |
| 12:48.39 | *** part/#brlcad jasvir (~jasvir@122.173.150.75) |
| 12:49.10 | ``Erik | brlcad, andromeda-galaxy: the new ray generation stuff, is that all being done as one big lump (here are your 2 million rays in one big honkin' array)? I'm wondering if a streaming generator might be handy? gen=init(), give_me_next_ray(gen), give_me_next_N_rays(gen,10,&array), etc? |
| 12:50.37 | ``Erik | raptor: for rt, -p <degrees> |
| 12:51.52 | Notify | 03BRL-CAD Wiki:Pavanaries * 0 /wiki/User:Pavanaries: |
| 13:05.16 | Notify | 02GCI:brlcad * 6632135641268224 : *code* comments :) - Hehe, Raptor, I meant comments *in* the code. Code // comments. So when someone reads these scripts in 5 years, they won't have to decipher... |
| 13:05.16 | Notify | 02GCI:brlcad * 5792678377684992 : cool, please communicate this with starseeker - So he can work with the fix. |
| 13:05.16 | Notify | 02GCI:brlcad * 5028787255246848 : Task Needs More Work - One of the mentors has sent this task back for more work. Talk to the mentor(s) assigned to this task to satisfy the requirements needed to... |
| 13:05.16 | Notify | 02GCI:brlcad * 5028787255246848 : almost - Krishna, this looks good to me. The only problem I saw is that you named the function argument "bu_n_free" and another variable "bu_n_free_calls". Using... |
| 13:05.16 | Notify | 02GCI:brlcad * 5581079297654784 : Task Needs More Work - One of the mentors has sent this task back for more work. Talk to the mentor(s) assigned to this task to satisfy the requirements needed to... |
| 13:05.16 | Notify | 02GCI:brlcad * 5581079297654784 : missing one? - Aditya, this looks nice but are you missing one? There are gaps in the grid and I only see 9 listed. Also, can you use the actual design from Marc... |
| 13:05.16 | Notify | 02GCI:goobypls * 5269456200663040 : Claim Removed - The claim on this task has been removed, someone else can claim it now. |
| 13:05.16 | Notify | 02GCI:goobypls * 5269456200663040 : Task Claimed - I would like to work on this task. |
| 13:05.16 | Notify | 02GCI:joy818200 * 6384883232407552 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:zhang_yizhuo * 4992155881832448 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:o7p9bxbnyj * 6703555041820672 : Task Claimed - I would like to work on this task. |
| 13:05.16 | Notify | 02GCI:goobypls * 5269456200663040 : Claim Removed - The claim on this task has been removed, someone else can claim it now. |
| 13:05.16 | Notify | 02GCI:o7p9bxbnyj * 6703555041820672 : Claim Removed - The claim on this task has been removed, someone else can claim it now. |
| 13:05.16 | Notify | 02GCI:o7p9bxbnyj * 6348864562921472 : Task Claimed - I would like to work on this task. |
| 13:05.16 | Notify | 02GCI:arclith * 5859273187786752 : Task Claimed - I would like to work on this task. |
| 13:05.16 | Notify | 02GCI:harman052 * 5859273187786752 : Task Assigned - This task has been assigned to Mou Yan Qiao. You have 100 hours to complete this task, good luck! |
| 13:05.16 | Notify | 02GCI:emiwang * 4949239830413312 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:adityagulati * 5581079297654784 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:adityagulati * 5581079297654784 : Explanation - Sir, One of the task (#6) has not been completed yet, So I could only make 9. I Have Tried To Make The rest Of The changes. Hope you like my... |
| 13:05.16 | Notify | 02GCI:adityagulati * 5581079297654784 : Explanation - Sir, Sorry I Could Not Increase The Size Much. Regards, Aditya |
| 13:05.16 | Notify | 02GCI:estafador * 5228725012529152 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:wenxuan45 * 5042004748664832 : Task Claimed - I would like to work on this task. |
| 13:05.16 | Notify | 02GCI:harman052 * 5042004748664832 : Task Assigned - This task has been assigned to Teoh Wen Xuan. You have 100 hours to complete this task, good luck! |
| 13:05.16 | Notify | 02GCI:tannousmarc * 6733016101552128 : Ready for review - The work on this task is ready to be reviewed. |
| 13:05.16 | Notify | 02GCI:tannousmarc * 6733016101552128 : Improvements, as discussed with Sean over IRC - - Now opens all the webpages on click, instead of linking to the task, for ease of use. - All the webpages... |
| 13:05.16 | Notify | 02GCI:wenxuan45 * 5042004748664832 : Claim Removed - The claim on this task has been removed, someone else can claim it now. |
| 13:05.16 | Notify | 02GCI:wenxuan45 * 4609906124521472 : Task Claimed - I would like to work on this task. |
| 13:05.17 | Notify | 02GCI:mihaineacsu29 * 6348864562921472 : Task Assigned - This task has been assigned to Andromeda Galaxy. You have 100 hours to complete this task, good luck! |
| 13:05.17 | Notify | 02GCI:princealikhan08 * 4596346644332544 : Task Claimed - I would like to work on this task. |
| 13:05.17 | Notify | 02GCI:princealikhan08 * 4596346644332544 : Work - Basically i have to create a logo for Google Code in 2015. is there any reference where i can get theme idea. |
| 13:05.17 | Notify | 02GCI:thevk * 4977508348854272 : Ready for review - The work on this task is ready to be reviewed. |
| 13:37.05 | raptor | I have already prepared the following 2 work :) |
| 13:50.10 | *** join/#brlcad mandarj (75c331dc@gateway/web/freenode/ip.117.195.49.220) |
| 13:50.35 | mandarj | gcibot, random brlcad |
| 13:50.48 | *** part/#brlcad mandarj (75c331dc@gateway/web/freenode/ip.117.195.49.220) |
| 14:03.50 | *** join/#brlcad YatharthROCK (~YatharthR@unaffiliated/yatharthrock) |
| 14:03.50 | *** part/#brlcad YatharthROCK (~YatharthR@unaffiliated/yatharthrock) |
| 14:27.52 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 14:37.25 | *** join/#brlcad jasvir (~jasvir@122.173.238.30) |
| 14:46.13 | *** join/#brlcad luca79 (~luca@host246-13-dynamic.0-87-r.retail.telecomitalia.it) |
| 14:52.04 | andromeda-galaxy | ``Erik: for now, as far as I know, it's returning an array of preallocated rays in a struct xrays, but that does make sense, since it's more like what rt does now... better yet, something like using give_me_nth_ray() might even be best...m |
| 14:52.23 | andromeda-galaxy | starseeker: if you were working with the comgeom modelling, did you see the fix to the scripts I made yesterday? |
| 14:53.29 | andromeda-galaxy | brlcad, ``Erik: then the way that do_pixel() works wouldn't have to change nearly as much, and there wouldn't be a huge memory allocations for all the rays... |
| 15:31.38 | andromeda-galaxy | <PROTECTED> |
| 15:35.48 | *** join/#brlcad teepee (~teepee@gateway/tor-sasl/teepee) |
| 15:55.51 | starseeker | you mjight want to think about something like give me next N rays, since one of the points with ray bundling is to handle blocks of rays "all at once" (in a sense) |
| 15:56.15 | starseeker | andromeda-galaxy: I hadn't seen the comgeom-g yet - so it was a script fix? awesome! |
| 15:56.38 | starseeker | will have to get that hooked up into the db builds |
| 16:07.33 | andromeda-galaxy | starseeker: makes sense... I was just thinking that for do_pixel() having random access is nice |
| 16:07.36 | andromeda-galaxy | but |
| 16:08.01 | andromeda-galaxy | starseeker: rt_shootrays() &c, could iterate through them using give_me_next_ray() similar to the way that they do with next() on the linked list |
| 16:08.05 | andromeda-galaxy | currently used for xrays |
| 16:10.55 | andromeda-galaxy | having give_me_nth_ray_range() might make sense as wellf for do_pixel() with hypersampling... |
| 16:12.58 | andromeda-galaxy | starseeker: yeah, the problem with it was that the script was making the region ident table in v1/v4 format, it had to move a couple of columns around to make it v5... |
| 16:26.05 | *** join/#brlcad MarcTannous (bc1b5e02@gateway/web/freenode/ip.188.27.94.2) |
| 16:26.16 | MarcTannous | Hello guys |
| 16:30.34 | MarcTannous | http://www.google-melange.com/gci/task/view/google/gci2014/6733016101552128 in case someone has 2 minutes |
| 16:30.34 | gcibot | [[ Create a web gallery of GCI website landing page designs || BRL-CAD || NeedsReview (2 days 5 hrs 37 min) || Sean, Dishank ]] |
| 16:45.03 | YashM | just got a google cardboard |
| 16:45.21 | YashM | this is cool :P |
| 16:51.14 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 16:55.33 | *** join/#brlcad luca79 (~luca@net-2-34-212-36.cust.vodafonedsl.it) |
| 17:03.41 | andromeda-galaxy | brlcad: thinking about ray bundle generation mechanisms, there seem to currently be two, one of which (older) stores rays in a bu_list and the other (newer) stores rays in an array of xray --- the second one seems better for rt due to the performance of random access in linked lists |
| 17:04.00 | andromeda-galaxy | but I'm not convinced that using a generator-like approach like ``Erik mentioned above isn't the best idea |
| 17:04.25 | andromeda-galaxy | something like a set of functions that operate on a struct like: |
| 17:04.30 | andromeda-galaxy | struct bundle_gen { |
| 17:04.39 | andromeda-galaxy | <PROTECTED> |
| 17:04.43 | andromeda-galaxy | <PROTECTED> |
| 17:05.06 | andromeda-galaxy | <PROTECTED> |
| 17:05.08 | andromeda-galaxy | } |
| 17:05.16 | andromeda-galaxy | with gen_func a function pointer to generation functionis |
| 17:06.42 | andromeda-galaxy | this has fast random access and alolcates memory lazily... |
| 17:13.55 | *** join/#brlcad merzo (~merzo@121-34-132-95.pool.ukrtel.net) |
| 17:26.59 | MarcTannous | yashm: Depending on the content of your div, text-align:center; or magin-left:auto;margin-right:auto; |
| 17:27.07 | MarcTannous | To center an element horizontally inside its parent div |
| 17:51.50 | raptor | Tell me why throws when creating sketch put test sketch V {0 0 0} A {1 0 0} B {0 0 1} VL { {-75 0 0} {75 0 0} {75 50 0} {75 -50 0} {-75 50 0} {-75 -50 0} } SL { {cars S 0 E 0 R 25 L 0 O 1} {cars S 1 E 1 R 25 L 0 O 1} {cars S 2 E 3 R 50 L 0 O 1} {cars S 4 E 5 R 50 L 0 O 1} {cars S 4 E 2 R 100 L 0 O 1} {cars S 5 E 3 R 100 L 0 O 1}} |
| 18:04.42 | raptor | in Archer |
| 18:05.35 | raptor | and mged |
| 18:07.56 | raptor | Thank you all, I found an error |
| 18:39.40 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 19:08.45 | starseeker | andromeda-galaxy: we want to make sure that whatever mechanisms we put in place for bunding allow us to implement things like GPU based raytracing and other coherence tricks |
| 19:09.01 | starseeker | andromeda-galaxy: I don't know much about the details - that's brlcad |
| 19:09.21 | starseeker | but I do know there are specific needs for high performance that aren't always intuitive |
| 19:09.26 | ``Erik | linked list shouldn't be an issue for 'normal' rendering, each worker is just "lock, pop, unlock" |
| 19:09.48 | ``Erik | if the ray info includes the x,y of the final pixel, then order can be changed to get 'postage stamp' bundles |
| 19:09.53 | starseeker | (not saying one approach or another won't succeed - I don't know - but I do know there's a lot of somewhat non-obvious factors that can enter in) |
| 19:10.38 | ``Erik | streaming them (and generating on demand) could be useful if, say, you wanted a 3 story tall poster at 1200dpi... :D or a normal poster at 120000 dpi or something |
| 19:12.02 | ``Erik | <-- is just spewing ideas, not arguing how it should be :D just hoping some interesting discussion might happen |
| 19:12.37 | andromeda-galaxy | ``Erik: you probably know more about this than I do, but here's hte question I have with linked lists: |
| 19:13.02 | andromeda-galaxy | each worker thread might need to access non-contiguous sections of the list of rays, right? Especially with the recursive refinement strategy... |
| 19:13.28 | andromeda-galaxy | but singly-linked lists are quite ineficient at random access... and rt is usually one of the more performance-critical sections of code, right? |
| 19:14.39 | andromeda-galaxy | ``Erik: so it seems like it makes more sense to at least have them layed out in a contiguous C array in memory or some such, but rt_shootrays() uses struct xrays, which uses bu_list... |
| 19:15.20 | ``Erik | well, my thought was that random access is irrelevant, but I'd forgotten about the -i option... to further muddy things up, the generator could be aware of -i and generate the rays to consider in the right order, possibly including a pixel "size" (-i level)? |
| 19:17.36 | ``Erik | rt is currently very built into doing things in a linear fashion, each worker does one entire line... "postage stamp" raytracing makes squares to raytrace to hopefully increase the probability that the geometry is already in the cpu cache |
| 19:17.47 | ``Erik | isst/adrt does postage stamps for that reason |
| 19:18.34 | ``Erik | I'd imagine any real benefits from contiguous memory would require significant retooling of the rt worker system and the icv interface... |
| 19:18.55 | andromeda-galaxy | ahh... |
| 19:19.18 | ``Erik | beyond the scope of this task... so, uh, I guess the best approach right now is to not worry about the speed of access |
| 19:19.19 | andromeda-galaxy | right, that makes sense... It's just that with n worker thread,s |
| 19:19.38 | andromeda-galaxy | they're accessing interleaved bits of memory, right? which might be pretty inneficient with a linked list... |
| 19:20.32 | Stragus | The coordinates of primary rays should be generated on the fly rather than stored in big buffers |
| 19:20.40 | andromeda-galaxy | ``Erik: so for this task, I can just write the generator without considering that, |
| 19:20.47 | andromeda-galaxy | but there's another one for making rt use them.. |
| 19:20.50 | ``Erik | the memory access patterns are pretty well tuned... for a vax 11/780 (and pretty solid on a mips r8k sgi origin)... for any cpu's BRL-CAD actually runs on right now, it's probably not so good |
| 19:20.54 | Stragus | And the coordinates of secondary rays should also be generated on the fly from whatever code handles the result of primary rays |
| 19:21.59 | andromeda-galaxy | ``Erik: what is the real benefit of using bundled rays like this instead of keeping rt's current on-the-fly generation? |
| 19:22.35 | andromeda-galaxy | Stragus: that was the rational behind the 'generator' idea, I believe |
| 19:23.06 | andromeda-galaxy | ``Erik: using a generator could be really interesting, because in some cases (i.e. current rt), it can delay the compuation until needed |
| 19:23.30 | andromeda-galaxy | but if it needs a block of, say, a hundred rays, it could just call, e.g., rt_bundle_next_n_rays(100) and get either a list or a buffer... |
| 19:24.17 | andromeda-galaxy | ``Erik: so the problem is that making rt use this kind of ray tracing might be quite difficult, especially re-doing that kind of memory optimization with the new structure, right? doing it this way also needs more malloc()s... |
| 19:27.39 | andromeda-galaxy | Stragus, ``Erik: are you still here? |
| 19:31.19 | andromeda-galaxy | ``Erik: My point about contiguous memory was just that random access from the different worker threads would (hopefully) be cheap |
| 19:33.38 | andromeda-galaxy | ``Erik, Stragus: it looks to me like the pixels are processed on n processors as 1234...n1234..n, which is my concern about using linked lists, since e.g. processor 1 has to follow n extra links... |
| 19:40.33 | Stragus | There are many inefficiencies in BRL-CAD's raytracer |
| 19:40.58 | Stragus | I wrote an efficient raytracer, both for CPUs and CUDA GPUs, but it's triangle-only |
| 19:41.26 | Stragus | BRL-CAD's CSG raytracer needs a major performance update |
| 19:49.32 | andromeda-galaxy | Stragus: ahh, interesting... so are you saying that it's already inefficient enough that this won't make matters much worse? |
| 20:02.38 | ``Erik | andromeda-galaxy: it's a bit more complicated than you think... linked lists are pretty fast to traverse, allocation is done using memory pools (no context switches, mru cache behaviors, etc) |
| 20:03.27 | ``Erik | logs into melange and tries to find the task description O.o |
| 20:05.01 | andromeda-galaxy | ``Erik: ah, good to know then.. mayabe it'll be fine |
| 20:05.19 | andromeda-galaxy | and about allocation, I meant allocating the xrayp structs, which is (currently in the other bundles, at least) being done with malloc |
| 20:09.58 | ``Erik | andromeda-galaxy: it'll be fine. :) beware premature optimization O.o |
| 20:20.51 | andromeda-galaxy | ``Erik: true... alright, then, I'll just finish it using the linked list for now, if it causes unacceptable performance degradations, we'll worry about it then. |
| 20:27.12 | *** join/#brlcad raptor_ (~raptor@194.28.75.221) |
| 20:30.24 | Stragus | ``Erik, linked lists are *terrible* for caches |
| 20:30.36 | Stragus | They are not fast to traverse by any mean |
| 20:31.35 | Stragus | It's better if the list is small enough (and allocated from memory pools) that it fits in L2 cache, but it's still not terrible |
| 20:41.11 | raptor_ | Is there anything like that on linux? http://brlcad.org/wiki/2D_Sketch_Editor |
| 20:42.19 | raptor_ | or windows :) |
| 21:17.14 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 21:21.15 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) |
| 22:03.29 | *** join/#brlcad raptor_ (~raptor@194.28.75.221) |
| 22:04.28 | *** join/#brlcad merzo (~merzo@121-34-132-95.pool.ukrtel.net) |
| 22:29.50 | *** join/#brlcad krishna_732 (4a0f1796@gateway/web/freenode/ip.74.15.23.150) |
| 23:22.38 | *** join/#brlcad raptor_ (~raptor@194.28.75.221) |
| 23:30.32 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 23:47.23 | *** join/#brlcad circ-user-tFqUU (~circuser-@122.163.139.152) |
| 23:47.34 | circ-user-tFqUU | Hi |
| 23:51.28 | *** join/#brlcad circ-user-2i4j3 (~circuser-@122.163.139.152) |
| 23:51.34 | circ-user-2i4j3 | Hi |
| 23:52.15 | circ-user-2i4j3 | I have claimed this task: http://www.google-melange.com/gci/task/view/google/gci2014/4635701832318976 |
| 23:52.15 | gcibot | [[ Create a web gallery of GCI website landing page designs #2 || BRL-CAD || Claimed (4 days 3 hrs 50 min) || Gauravjeet Singh, Ch3ck ]] |
| 23:53.23 | circ-user-2i4j3 | I cannot find any completed "landing" tasks on the BRL-CAD's page. Please look into it. |
