| 10:29.42 | *** join/#brlcad infobot (ibot@rikers.org) |
| 10:29.42 | *** 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. |
| 11:16.53 | MarcTannous | So little activity on a Sunday :O |
| 11:17.21 | Notify | 03BRL-CAD Wiki:Utkarshsingh * 0 /wiki/User:Utkarshsingh: |
| 11:24.09 | YashM_ | btw today is BRL-CAD's 10th open source anniversary |
| 11:38.22 | *** join/#brlcad rohitagarwal (~Rohit@49.249.53.179) |
| 11:45.11 | MarcTannous | brlcad: check this out http://codepen.io/marctannous/pen/xbOBGP, regarding our past JS canvas discussion. |
| 11:47.10 | MarcTannous | correct link : http://codepen.io/marctannous/pen/xbOBGP |
| 11:50.00 | MarcTannous | is this what you had in mind? |
| 11:58.03 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 12:01.43 | ignacio | hi all, good morning! :) |
| 12:02.16 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 12:14.30 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 12:17.17 | Notify | 02GCI:Melange * 6359310477033472 : Task Reopened - Melange has detected that the final deadline has passed and it has reopened the task. |
| 12:18.00 | Notify | 02GCI:mehak_s * 6359310477033472 : Task Claimed - I would like to work on this task. |
| 12:24.25 | Notify | 02GCI:erikg * 6359310477033472 : Task Assigned - This task has been assigned to Mehak Sharma. You have 100 hours to complete this task, good luck! |
| 12:26.56 | Notify | 02GCI:erikg * 6359310477033472 : Time spent is concerning - Mehak, you've been working on this for 3 weeks now with several extensions and re-assignments. This seems excessive. Don't be surprised... |
| 12:33.22 | ``Erik | cpu cache performance effects http://igoro.com/archive/gallery-of-processor-cache-effects/ (someone made a combined graph at https://dl.dropboxusercontent.com/u/4893/mem_lat3.jpg ) |
| 12:41.19 | Notify | 02GCI:mehak_s * 6359310477033472 : None - Okay sir, Sorry for the inconvenience caused. I promise this time I will complete the task without any extensions or re-assessments .. |
| 12:41.54 | Notify | 02GCI:Melange * 6094206841913344 : Task Reopened - Melange has detected that the final deadline has passed and it has reopened the task. |
| 12:42.59 | Notify | 02GCI:mehak_s * 6359310477033472 : Ready for review - The work on this task is ready to be reviewed. |
| 12:44.13 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 12:47.56 | Notify | 02GCI:helix * 6094206841913344 : Task Claimed - I would like to work on this task. |
| 12:49.58 | *** join/#brlcad mehak_s (~sarthak@59.89.98.239) |
| 12:51.04 | mehak_s | Erik sir? |
| 12:51.52 | mehak_s | could you please check my task? |
| 12:52.29 | mehak_s | http://www.google-melange.com/gci/task/view/google/gci2014/6359310477033472 |
| 12:52.31 | gcibot | Design a banner for our Facebook page (and/or other social media sites) #3 |
| 12:52.31 | gcibot | Status: NeedsReview (4 days 3 hrs 31 min) |
| 12:52.31 | gcibot | Mentor(s): Isaac Kamga, Sean |
| 12:55.39 | Notify | 02GCI:Melange * 4639899508539392 : Task Reopened - Melange has detected that the final deadline has passed and it has reopened the task. |
| 13:05.24 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 13:15.49 | *** join/#brlcad mehak_s (~sarthak@117.200.54.17) |
| 13:39.07 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 13:44.23 | Notify | 02GCI:shubham2810 * 5232150064398336 : Task Claimed - I would like to work on this task. |
| 14:14.52 | Notify | 02GCI:gjeet * 6094206841913344 : Task Assigned - This task has been assigned to Yash. You have 100 hours to complete this task, good luck! |
| 14:15.08 | Notify | 02GCI:gjeet * 5232150064398336 : Task Assigned - This task has been assigned to Shubham Aggarwal. You have 100 hours to complete this task, good luck! |
| 14:21.20 | Notify | 02GCI:gjeet * 6359310477033472 Design a banner for our Facebook page (and/or other social media sites) #3 https://www.google-melange.com/gci/task/view/google/gci2014/6359310477033472: Nice and... |
| 14:21.40 | Notify | 02GCI:gjeet * 6359310477033472 : 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... |
| 14:29.19 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 14:39.04 | Notify | 02GCI:prachipahuja * 6426124666011648 : Ready for review - The work on this task is ready to be reviewed. |
| 15:08.05 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 15:23.48 | YashM_ | soooo quiet |
| 15:29.35 | Notify | 02GCI:popescuandrei * 5900674122383360 : None - Marc, you need to fix your system so you can build. What's stopping you from doing so ? Ask for help ! |
| 15:30.56 | Notify | 02GCI:mehak_s * 6359310477033472 : None - My friend made the background for me :) She is also a GCI participant.. :) |
| 15:32.31 | Notify | 02GCI:popescuandrei * 5900674122383360 : 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... |
| 15:32.32 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 15:32.46 | Notify | 02GCI:mehak_s * 6359310477033472 : None - And also it is the only fond through which text is a little readable lol |
| 15:39.07 | *** join/#brlcad mehak_s (~sarthak@117.200.57.237) |
| 15:49.32 | Notify | 02GCI:gjeet * 6359310477033472 : Hmm - Ok then mark the task as complete again so that I can approve it. |
| 15:55.05 | starseeker | andromed1-galaxy: nice work with the NACA tasks! So it looks like we need another task or two for verification? |
| 15:55.14 | *** join/#brlcad gjeet (75d43345@gateway/web/cgi-irc/kiwiirc.com/ip.117.212.51.69) |
| 15:55.48 | starseeker | brlcad: how do you want to handle the build client setup for buildbot/cdash? Can we set up a couple small dedicated purpose virtual machines on bz? |
| 16:00.24 | Notify | 02GCI:mehak_s * 6359310477033472 : Ready for review - The work on this task is ready to be reviewed. |
| 16:01.47 | starseeker | that won't help for OSX or Windows, but we could get flavors of Linux and BSD, plus Haiku and (if we're feeling brave) OpenIndiana |
| 16:02.14 | starseeker | well, I suppose we could do Windows but we'd have to get a license... OSX is a problem |
| 16:06.15 | Notify | 02GCI:gjeet * 6359310477033472 : Task Closed - Congratulations, this task has been completed successfully. |
| 16:08.08 | ``Erik | vm's tend to clobber memory and bz isn't exactly a memory monster anymore... (we're running about 1gig in swap... one of the jvm's burns another gig) |
| 16:08.20 | starseeker | winces |
| 16:13.58 | *** join/#brlcad mehak_s (~sarthak@117.200.59.163) |
| 16:14.23 | starseeker | ``Erik: what about the GCC compile farm? |
| 16:15.47 | ``Erik | that might work... or other donors... I'd note that a now defunct project sent tons of data from arl machines to external servers, might be able to make a case for tagging a handful of machines there as build clients? (just phrase it in the appropriate enterprisey way) |
| 16:16.03 | ``Erik | now defunct java project, even, if'n ya know what I mena |
| 16:16.37 | ``Erik | cdash build results are XML so they can easily be audited |
| 16:17.44 | Notify | 02GCI:azadanh * 4956557716488192 : Task Claimed - I would like to work on this task. |
| 16:18.54 | Notify | 02GCI:gjeet * 4956557716488192 : Task Assigned - This task has been assigned to AzadanH. You have 100 hours to complete this task, good luck! |
| 16:23.13 | Notify | 02GCI:popescuandrei * 5216595353796608 : Task Closed - Congratulations, this task has been completed successfully. |
| 16:25.58 | *** join/#brlcad alisha (~alisha@101.58.163.37) |
| 16:30.07 | *** join/#brlcad andrei_ (~androirc@86.124.72.236) |
| 16:30.14 | andrei_ | Hello |
| 16:42.18 | *** join/#brlcad mehak_s (~sarthak@117.200.59.163) |
| 16:44.51 | Notify | 02GCI:azadanh * 4956557716488192 : Claim Removed - The claim on this task has been removed, someone else can claim it now. |
| 16:54.25 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 16:55.35 | *** join/#brlcad alisha (~alisha@101.60.198.13) |
| 17:22.30 | *** join/#brlcad mihaineacsu (~mihaineac@92.85.30.25) |
| 17:28.12 | Notify | 02GCI:maths22_m * 6426124666011648 : Looks good - It looks like you got BRL-CAD up and running, and were able to successfully create your first model. Congratulations! |
| 17:28.17 | Notify | 02GCI:maths22_m * 6426124666011648 : Task Closed - Congratulations, this task has been completed successfully. |
| 17:28.18 | Notify | 02GCI:maths22_m * 6426124666011648 : Task Closed - Congratulations, this task has been completed successfully. |
| 17:33.42 | maths22 | starseeker, ``Erik: What do we think about installing cmake on bz from a bootstrap build? |
| 17:39.28 | *** join/#brlcad mehak_s (~sarthak@117.200.49.38) |
| 17:47.59 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 18:03.11 | Notify | 02GCI:tannousmarc * 5900674122383360 : Conclusion - There seems to be missing a cmake file to actually "make test" on, so I'll reinstall in the morning, however that will take a lot of time and I'd... |
| 18:03.16 | Notify | 02GCI:tannousmarc * 5900674122383360 : Ready for review - The work on this task is ready to be reviewed. |
| 18:15.32 | Notify | 02GCI:martinhesa * 6187982620983296 : Task Claimed - I would like to work on this task. |
| 18:16.12 | Notify | 02GCI: * : - |
| 18:26.55 | starseeker | maths22: works for me |
| 18:27.19 | starseeker | ``Erik and brlcad are the ones who'll have to make that call though |
| 18:37.48 | maths22 | starseeker: Take a look: http://brlcad.org/CDash/index.php?project=BRL-CAD&date=2014-12-21 |
| 18:38.01 | maths22 | I used my ctest and got it to happily upload |
| 18:38.22 | MarcTannous | maths22: Could you please take a look over my task? |
| 18:38.49 | MarcTannous | http://www.google-melange.com/gci/task/view/google/gci2014/5900674122383360 I've done all rossberg wanted to change and the only impediment is me not being able to run tests as I'd have to reinstall rt^3 which I do not have the time to |
| 18:38.50 | gcibot | Write CoreInterface unit test #3 |
| 18:38.50 | gcibot | Status: NeedsReview (46 hrs 48 min) |
| 18:38.50 | gcibot | Mentor(s): Sean, Ishwerdas |
| 18:38.53 | MarcTannous | But my code is there :( |
| 18:39.03 | starseeker | maths22: nice! |
| 18:39.46 | starseeker | maths22: I'm certainly not attached to the included CMake - I'm curious why it fails, but if bootscrapping gets us running I say go for it |
| 19:54.12 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 20:03.14 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 20:09.19 | MarcTannous | has anyone seen this? http://learningthreejs.com/blog/2013/08/02/how-to-do-a-procedural-city-in-100lines/ |
| 20:24.23 | Notify | 02GCI:ankitmukherjee * 6359310477033472 : Decision - Are you going to send the certificate of completion to the person who has their design is chosen or . . . ? I am a little confused as to how... |
| 20:28.55 | andromed1-galaxy | I'm working on http://www.google-melange.com/gci/task/view/google/gci2014/5860933276532736, and I have the monotone chain code apparently creating convex hulls; however, something else seems to be causing it to crash later... specifically, on line 2227, when "ProjPointSet pps;" isn't executed it doesn't segfault... anyone have an idea as to why that might be? The output of the algorithm appears to be correct... |
| 20:29.39 | starseeker | hmm |
| 20:30.20 | starseeker | I'll check - I guess the first question is what is the difference between the output from the monotone chain as compared to the older code? (assuming the older code doesn't crash with this input?) |
| 20:30.38 | andromed1-galaxy | I think it's mostly in the ordering of the points |
| 20:30.48 | starseeker | hmm |
| 20:32.22 | starseeker | does the older code appear to be sorting them in any way? |
| 20:32.26 | andromed1-galaxy | actually, looking it over again, it seems like points that are duplicated in the input set sometimes come out duplicated in the result... I don't know if that could be a problem or not... |
| 20:32.36 | starseeker | possibly |
| 20:32.49 | andromed1-galaxy | I don't think it's sorting them in any particular way... |
| 20:32.59 | andromed1-galaxy | the other strange thing is that this code seems to run ~100 times before crashing |
| 20:33.09 | starseeker | O.o |
| 20:33.49 | andromed1-galaxy | the old code is run ~600 times ina row |
| 20:34.25 | starseeker | the quickest test I can think to make is to take the monotone chain output and create a "new" output which makes sure any duplicate points in the input are duplicated in that output |
| 20:34.41 | andromed1-galaxy | sorry? |
| 20:35.05 | andromed1-galaxy | the monotone chaian algorithm has duplicates in the result, I'm not sure whether the graham scana did or not |
| 20:35.13 | starseeker | oh! |
| 20:35.35 | starseeker | then maybe pre-process the input to eliminate duplicates? |
| 20:35.45 | andromed1-galaxy | (I'm trying to find things in the files of output as we speak, but they're quite large, with diagnostic output from ~6000 runs in each) |
| 20:35.51 | starseeker | try feeding it to the old and the new without duplicates... |
| 20:35.58 | starseeker | winces |
| 20:36.42 | starseeker | andromed1-galaxy: if it gets too messy, there's another possibility - ditch this code and try to get the geometricsolutions version working |
| 20:37.08 | andromed1-galaxy | true... |
| 20:37.15 | starseeker | that has its own quirks, but it should at least be somewhat cleaner |
| 20:37.18 | andromed1-galaxy | in at least one case near the end, the graham scan did have some of the points duplicated... |
| 20:38.54 | starseeker | andromed1-galaxy: my modus operandi in that situation would be to try to make the two outputs look as similar as possible to see what difference or differences trigger the trouble |
| 20:40.00 | andromed1-galaxy | indeed.... it's weird, for at least the first few cases they look pretty much the same other than ordering (which can't really be avoided since the monotone chain finds points in a different order than graham scan) |
| 20:40.24 | starseeker | andromed1-galaxy: if it gets too long as a task, go ahead and report back what has been done so far and we can consider what approach to hit next |
| 20:41.24 | starseeker | if the code looks like a real pain to get working well, we probably should ditch it for the geometric solutions implementation (at least for a test) especially since the license is better on the gs code |
| 20:41.37 | andromed1-galaxy | starseeker: sure... I may need to do that fairly soon, I've already been looking at it for ~2:45 |
| 20:41.48 | starseeker | nods |
| 20:41.54 | andromed1-galaxy | do you mean all of gdiam, or just the monotone chain? |
| 20:42.00 | starseeker | all of gdiam |
| 20:42.10 | starseeker | it's only being used for the oriented bounding box code |
| 20:42.16 | andromed1-galaxy | ah, that makes sense... |
| 20:42.54 | starseeker | er, Geometric Tools rather |
| 20:43.28 | starseeker | http://www.geometrictools.com/Samples/Geometrics.html#MinimumVolumeBox3D specifically |
| 20:44.24 | andromed1-galaxy | ahh... that looks like it might be good to use... |
| 20:45.06 | andromed1-galaxy | well, I'll look at gdiam a little more and then submit this with my findings... If I can't get it working, maybe we can do another task to try to switch to the geometrictools code... |
| 20:45.12 | starseeker | that will require some subset of their code (not sure how much) be integrated under libbn. Minimally it will require their convex hull algorithm, which will be a useful exercise since we can compare it to clarkson's |
| 20:45.19 | starseeker | andromed1-galaxy: sounds good |
| 20:45.48 | andromed1-galaxy | for this one, I'll submit the debugging output files I made as well as the code changes with the debugging in them, though I'm not sure how useful those will be... |
| 20:45.49 | starseeker | andromed1-galaxy: the geometric tools code is C++11, which is too new - we'd need to step it down, although not all the way to C |
| 20:45.56 | andromed1-galaxy | sounds interesting, I look forward to trying it! |
| 20:46.01 | starseeker | andromed1-galaxy: sounds good |
| 20:46.42 | starseeker | don't give up yet - you may find everything works if you make the monotone chain code output closer to the old output |
| 20:47.26 | starseeker | in principle the specific routine for the convex hull calculation should make a difference only from a robustness standpoint, unless there is some point ordering needed by some other part of the algorithm |
| 20:47.55 | andromed1-galaxy | true... like I said, the output appear to be pretty much the same except for ordering, and I'm really not sure how to make the ordering the same... |
| 20:48.09 | starseeker | yeah, that's what's worrying me |
| 20:48.28 | starseeker | I'd try the duplicate points thing, to make sure the outputs are the same size |
| 20:48.45 | starseeker | beyond that, if there really is an ordering dependency, that's going to be an issue |
| 20:50.01 | starseeker | will see if he can extract the subset of files needed for the mbr code, to make scoping the other approach easier |
| 20:50.50 | andromed1-galaxy | starseeker: makes sense.... I'll looki into it some more and see if I can find too many other differences... |
| 20:55.26 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 20:55.33 | MarcTannous | Good evening guys |
| 21:03.06 | starseeker | andromed1-galaxy: if you want to take a sneak peek, I think this is the code that they use to define the minimum box: brlcad.org/~starseeker/GteMinimumVolumeBox3.tar.gz |
| 21:04.06 | andromed1-galaxy | starseeker: geometrictools? |
| 21:04.08 | starseeker | there are several approaches - the best of all worlds would be to translate everything there into C and libbn structures, with a little C++ for things like set and map if we need log n behavior in containers |
| 21:04.12 | starseeker | yep |
| 21:04.25 | starseeker | around 7k lines |
| 21:04.51 | andromed1-galaxy | interesting... |
| 21:05.13 | starseeker | some of those we have (2d convex hull and bounding area) |
| 21:05.41 | andromed1-galaxy | right... libbn has a 2d convex hull, right? |
| 21:05.48 | starseeker | nods |
| 21:05.54 | starseeker | and a 2d minimum bounding rectangle |
| 21:06.25 | starseeker | nominally it has a 3d convex hull now with the clarkson code, but it would probably be worth trying to port the geometrictools implementation in |
| 21:06.37 | starseeker | if nothing else, hook it up and see how it does performance wise |
| 21:07.20 | andromed1-galaxy | right... how exactly do 2d/3d convex hulls connect to oriented bounding boxes? sorry if there's something basic I'm missing. |
| 21:07.38 | starseeker | it has to do with the stages of the algorithm |
| 21:08.31 | starseeker | the bounding box of the convex hull will be the same as the bounding box of the object itself (for polygonal meshes anyway) |
| 21:08.48 | starseeker | so you find the hull, which simplifies the actual calculation of the box |
| 21:09.10 | andromed1-galaxy | makes sense |
| 21:09.19 | andromed1-galaxy | how does gdiam manage to do it all with 2d hulls/ |
| 21:09.20 | andromed1-galaxy | ? |
| 21:09.38 | starseeker | it's in the paper linked to by the task |
| 21:09.49 | starseeker | I don't pretend to understand all the details of that algorithm |
| 21:10.03 | andromed1-galaxy | ah, okay... |
| 21:10.23 | starseeker | a third approach to the problem would be to read the paper and re-implement the algorithm in libbn ;-) |
| 21:10.33 | andromed1-galaxy | true :-) |
| 21:10.50 | starseeker | I'm assuming that's hard, given the size of the gdiam code |
| 21:11.13 | andromed1-galaxy | probably... |
| 21:11.20 | starseeker | the geometrictools implementation may be slower than the gdiam code, but I can live with that if it works reliably |
| 21:11.35 | starseeker | we can refine it with the gdiam approach someday |
| 21:12.09 | andromed1-galaxy | so, just re-running the old & new chull code with gdiam right now, the new code runs 1001 times and then something (not it) crashes, and the old ran a total of 6807 times (with the example from the task description) |
| 21:12.13 | andromed1-galaxy | indeed |
| 21:12.52 | starseeker | bugger |
| 21:13.01 | *** join/#brlcad chick_ (~capslock@41.205.22.41) |
| 21:13.06 | starseeker | and as near as you can tell, only the point ordering was different? |
| 21:14.19 | andromed1-galaxy | it's hard to tell for sure, because comparing all 1001 cases is somewhat difficult to say the least... I am (and have been) trying to take a (semi)-random sampling... |
| 21:14.47 | andromed1-galaxy | so far, that's been the case |
| 21:14.51 | starseeker | can you code it? Run both the old and new routine on each iteration and check if they're different? |
| 21:15.14 | starseeker | (in size) |
| 21:15.23 | *** join/#brlcad FreezingCold (~FreezingC@135.0.41.14) |
| 21:15.40 | andromed1-galaxy | might be able to... |
| 21:16.21 | andromed1-galaxy | right now, the two versions are in the middle of the function with one or the other #ifdef'd out, but I could probably make it work... do you know if the iterations depend on the result of the previous one? |
| 21:16.43 | andromed1-galaxy | starseeker: if so, then they must all be the same - ordering because the last one executed by the new algorithm matches the results from the old |
| 21:17.59 | andromed1-galaxy | aha! I managed to find a case where the new one has duplicates and the old doesn't... I wonder why... both are supposed to eliminate the duplicates before running the algorithm |
| 21:24.57 | *** join/#brlcad AndroUser2 (~androirc@86.121.196.213) |
| 21:30.58 | *** join/#brlcad ries (~ries@D979C47E.cm-3-2d.dynamic.ziggo.nl) |
| 21:33.28 | andromed1-galaxy | starseeker: in case you're still looking into the gdiam stuff, it looks like minmax is being set to some huge negative |
| 21:33.44 | andromed1-galaxy | value despite being initialized to i-1... |
| 21:34.29 | andromed1-galaxy | oh wait, never mind, something else was wrong with the debugging... |
| 21:41.16 | andromeda-galaxy | starseeker: okay, I think I've removed all duplicates that aren't also duplicates in the original one... I'll look into automatedly comparing results for a little while, then submit what I've got so far... |
| 21:45.10 | Notify | 02GCI:ppungchai * 4956557716488192 : Task Claimed - I would like to work on this task. |
| 22:36.27 | *** join/#brlcad FreezingCold (~FreezingC@135.0.41.14) |
| 23:09.32 | Notify | 02GCI:o7p9bxbnyj * 5860933276532736 : Ready for review - The work on this task is ready to be reviewed. |
| 23:20.30 | Notify | 02GCI:o7p9bxbnyj * 5860933276532736 : Diff files - Sorry, the first diff contains a couple of extra debugging checks that I was using to make sure that out-of-bounds errors didn't occur that aren't... |
| 23:23.10 | starseeker | andromeda-galaxy: might want to add a note for the reviewer about the switch not producing a working result... |
| 23:24.40 | starseeker | as a reminder that we need more tasks |
| 23:27.04 | maths22 | starseeker: Look here http://brlcad.org/CDash/buildSummary.php?buildid=5 |
| 23:27.07 | maths22 | Windows |
| 23:27.12 | starseeker | except... that seemed to work |
| 23:27.17 | starseeker | andromeda-galaxy: did it work?? |
| 23:27.31 | maths22 | Does anyone have an OS X platform to try it on? |
| 23:27.38 | maths22 | ctest -D Experimental |
| 23:28.27 | starseeker | maths22: nice! |
| 23:28.42 | starseeker | looks like we'll have to figure out how to have it avoid the stepcode output |
| 23:29.00 | starseeker | I think I can quiet that with CMake, actually - I'll try that first |
| 23:29.32 | andromeda-galaxy | starseeker: it suddenly stated working... |
| 23:30.23 | andromeda-galaxy | maths22: I don't, sorry... what were we using as buildslaves for the buildbot instance back when it was last up? |
| 23:30.55 | andromeda-galaxy | starseeker: I think that one of the changes that I made to remove the duplicated points made it start working... |
| 23:31.37 | starseeker | andromeda-galaxy: hot diggity! |
| 23:31.44 | starseeker | nice work! |
| 23:31.51 | andromeda-galaxy | starseeker: thanks1 |
| 23:31.54 | andromeda-galaxy | s/1/!/ |
| 23:32.21 | andromeda-galaxy | the box it makes in mged looks right to me, at least for the test model described in the task... |
| 23:32.26 | starseeker | nods |
| 23:32.44 | starseeker | the key will be to reproduce some of the earlier crashing cases and see if they still wipe out |
| 23:33.08 | starseeker | I may have some of those tucked away somewhere, but I'll have to dig |
| 23:33.15 | starseeker | IIRC, it was easier to make it wipe out on Windows |
| 23:33.50 | andromeda-galaxy | interesting... I didn't see anything that looked particularly platform-specific in the graham scan code |
| 23:33.51 | starseeker | I think the last time we had build slaves up it was individuals hooking in their machines |
| 23:34.27 | andromeda-galaxy | ah... I was just curious, because it looks like we had a decent set of OSes for testing on for the buildbot last time.. |
| 23:34.31 | starseeker | andromeda-galaxy: it had to do with a unique ordering comparision deep in the STL implementation, IIRC - a required guarantee for unique ordering was being checked, and was failing |
| 23:34.43 | starseeker | nods - I think those were some student machines :-) |
| 23:35.14 | andromeda-galaxy | ah, I see.. |
| 23:35.19 | starseeker | andromeda-galaxy: it was down to platform specific sensitivities to things like floating point noise and tolerancing whether or not the ordering stayed unique |
| 23:35.52 | andromeda-galaxy | right... that's annoying... I hope the monotone chain algorithm works more reliably |
| 23:36.00 | maths22 | starseeker: how do you plan to silence the stepcode stuff? |
| 23:36.05 | starseeker | at least one of the problems seemed tcome with the graham scan trying to calculate areas for verrry nearly degenerate triangles |
| 23:36.18 | maths22 | Also, cdash maxes out at 50 errors/warnings |
| 23:36.20 | starseeker | maths22: I'll just capture the standard output to a variable when running the converters |
| 23:37.15 | starseeker | we don't have to be verbose about it - I mainly left it to show the conversions were actually working, since it was the first time I was building a sample db file that isn't in the BRL-CAD asc format |
| 23:37.30 | andromeda-galaxy | strseeker: ah, right --- I saw that in one of the comments. That, at least, shouldn't be as much of a problem for the monotone chain code |
| 23:37.38 | starseeker | exactly |
| 23:37.58 | starseeker | same reason I used the code you were inserting there as the basis for the libbn implementation |
| 23:38.15 | starseeker | maths22: give me a sec - working it now |
| 23:39.21 | starseeker | hmm... well, I thought I could |
| 23:39.24 | andromeda-galaxy | starseeker: makes sense... well, hopefully this will be useful --- I'm glad it started working, anyway |
| 23:39.25 | starseeker | uh oh |
| 23:39.34 | starseeker | andromeda-galaxy: indeed - thank you! |
| 23:39.55 | starseeker | if it proves out in testing, we'll send it back upstream to the author as a 1.0.2 version :-) |
| 23:41.35 | andromeda-galaxy | starseeker: hope so! probably the largest difference was in accomodating std::vector for the output instead of <C array of size at least n> |
| 23:42.07 | starseeker | nods |
| 23:46.36 | Notify | 03BRL-CAD:starseeker * 63801 brlcad/trunk/db/nist/CMakeLists.txt: Redirect output from step conversions to log files instead of stdout. |
| 23:46.46 | starseeker | maths22: I think that'll work |
| 23:47.33 | starseeker | does on Linux, and I think CMake honors the output redirect on Windows (we'll have to check) |
| 23:48.16 | starseeker | maths22: can we run make check and make regress instead of make test, or does ctest not give use the flexibility when using the Dash logic? |
| 23:48.55 | maths22 | Maybe? |
| 23:48.56 | maths22 | Why? |
| 23:49.16 | starseeker | make test runs everything, including tests that are known not to work |
| 23:49.24 | starseeker | flawfinder, rtwizard, etc |
| 23:49.38 | starseeker | check and regress are the ones we've "hooked up" and expect to work |
| 23:50.06 | starseeker | if it's easier we can comment out the non-working ones |
| 23:50.18 | starseeker | is fine either way |
| 23:51.10 | starseeker | advantage of leaving them on is people can work with them in isolation "out of the box" if they want to try and get them working without having to uncomment code, but the nightly regression is more important |
| 23:52.04 | maths22 | starseeker: I can't find an easy way to change that |
| 23:52.15 | starseeker | maths22: OK, don't worry about it |
| 23:54.17 | starseeker | surprised the quat unit tests are failing |
| 23:54.43 | starseeker | checks on Linux |
| 23:55.00 | starseeker | blinks |
| 23:55.06 | starseeker | yeah, failing here too |
| 23:55.12 | starseeker | wonder what busted 'em |
