| 00:21.21 | starseeker | probably shouldn't need this but wants to read it anyway: http://www.amazon.com/Art-Debugging-GDB-DDD-Eclipse/dp/1593271743 |
| 00:47.36 | brlcad | starseeker: did the file have a trailing newline? |
| 00:47.45 | brlcad | er, newline+cr |
| 00:48.37 | brlcad | Ahh.. ddd ... the first debugger I ever used. |
| 00:48.51 | brlcad | that was a neat debugger, pretty memory graphs |
| 00:48.59 | Ralith | oo, it has graphs? |
| 00:49.01 | Ralith | install |
| 00:49.24 | brlcad | yeah, if you had a linked list, you'd see your memory nodes and how they link together |
| 00:49.59 | brlcad | very simple ones, but enough to get the idea of what the data is doing |
| 00:50.36 | starseeker | brlcad: it did, but that shouldn't matter |
| 00:50.40 | brlcad | http://v3.sk/~lkundrak/grub2-gdb/ddd-ls.png |
| 00:51.03 | brlcad | starseeker: so red is still busted on windows? |
| 00:51.14 | starseeker | yes - in fact it never worked |
| 00:51.47 | starseeker | wonders how many white hairs he has racked up due to red... |
| 00:52.18 | brlcad | 'never' is a very long time with brl-cad, you sure? :) |
| 00:52.59 | starseeker | well, the new regex based version never worked |
| 00:53.08 | brlcad | that I'd believe :) |
| 00:53.21 | CIA-14 | BRL-CAD: 03starseeker * r43825 10/brlcad/branches/cmake/src/libged/red.c: MFC r43824 |
| 00:53.28 | starseeker | the regex didn't account for the brain-dead windows approach to line endings, and once we got past that the EOF situation causes a crash |
| 00:54.23 | brlcad | that's right, you are using some regex extension to match lines right? |
| 00:54.36 | CIA-14 | BRL-CAD: 0386.163.157.31 07http://brlcad.org * r2482 10/wiki/Main_Page: /* Getting started */ |
| 00:54.37 | starseeker | match across multiple lines, yes |
| 00:55.26 | starseeker | it LOOKS like regexec is running right off the end of the bu_mapped_file buffer |
| 00:55.32 | Ralith | pretty |
| 00:55.43 | Ralith | and by pretty I mean useful |
| 00:55.55 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2483 10/wiki/Main_Page: Reverted edits by [[Special:Contributions/86.163.157.31|86.163.157.31]] ([[User talk:86.163.157.31|Talk]]); changed back to last version by [[User:Paulcs|Paulcs]] |
| 00:56.01 | brlcad | Ralith: :) |
| 00:56.16 | starseeker | wonders if he can convince work to pick up a copy of that book... |
| 00:56.29 | brlcad | ddd got its groove on back in the early 90's |
| 00:56.36 | brlcad | hasn't really ever changed unfortunately, lot of potential there |
| 00:57.08 | brlcad | starseeker: it very well may be running off the end if it's looking for an EOF |
| 00:57.19 | CIA-14 | BRL-CAD: 0386.163.157.31 07http://brlcad.org * r2484 10/wiki/Error_Messages: Initial stub |
| 00:57.27 | Ralith | suddenly SLIME doesn't hold all the cards anymore |
| 00:59.28 | starseeker | terminating with \0 may work too... that's what OSX shows me after the mapped file, anyway... |
| 01:00.03 | starseeker | had hoped the bu_mapped_file routines would have some kind of guarantee about sanity at the end of the buf, but maybe not |
| 01:04.26 | brlcad | it's just a buffer of data |
| 01:06.13 | starseeker | then I'm gonna have to do something else with it |
| 01:06.24 | starseeker | because on every platform but Windows, regexec knows to stop |
| 01:06.33 | brlcad | now what could be happening is that on windows, it doesn't have mmap() so bu_mapped_file() is falling back to simple write() calls |
| 01:06.54 | brlcad | and if it was off-by-one, it could leave off the trailing '\0' |
| 01:08.10 | brlcad | you'd probably not even notice that on code that processed one line at a time |
| 01:11.21 | starseeker | looks like it used read() to pull it in |
| 01:13.23 | CIA-14 | BRL-CAD: 03brlcad * r43826 10/brlcad/trunk/src/libbu/mappedfile.c: make sure the buffers zero-terminate in case the files being mapped do not |
| 01:14.09 | brlcad | don't know if that'll fix it, but easy enough to test |
| 01:14.16 | starseeker | is on it |
| 01:16.19 | CIA-14 | BRL-CAD: 03starseeker * r43827 10/brlcad/branches/cmake/src/libbu/mappedfile.c: MFC r43826 |
| 01:18.28 | starseeker | yep, that's got it |
| 01:18.33 | starseeker | brlcad: thanks!! |
| 01:18.59 | starseeker | will have to check for other cases were n isn't enough in regex expressions |
| 01:20.42 | starseeker | looks like red may be the only case... |
| 01:21.31 | brlcad | your regex doesn't look right |
| 01:21.48 | brlcad | [[:blank:]\r?\n] |
| 01:22.08 | brlcad | presumably implying \r is optional, but that's not what ? means there |
| 01:22.21 | brlcad | it's a char class |
| 01:22.30 | brlcad | [[:blank:]\r\n]* |
| 01:22.40 | brlcad | or [[:blank:]\r\n]+ |
| 01:22.56 | starseeker | hmm... wonder why it worked |
| 01:23.08 | brlcad | because you're matching a literal '? |
| 01:23.16 | brlcad | exactly once |
| 01:23.32 | brlcad | ? or space or \r or \n |
| 01:24.14 | brlcad | several places that pattern repeats |
| 01:25.32 | starseeker | ah |
| 01:25.41 | starseeker | so I probably don't need the ? then |
| 01:27.35 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r0 10/wiki/Special:Log/delete: |
| 01:27.35 | CIA-14 | BRL-CAD: deleted "[[Error Messages]]": there needs to be more purpose than just itemizing |
| 01:27.35 | CIA-14 | BRL-CAD: possible error messages coming from brl-cad tools. the one you listed isn't |
| 01:27.35 | CIA-14 | BRL-CAD: even exactly an 'error' as it is an explanation why it skips _GLOBAL |
| 01:30.54 | CIA-14 | BRL-CAD: 03starseeker * r43828 10/brlcad/trunk/src/libged/red.c: Tweak the regex expressions for carriage return correctness - hopefully that's closer |
| 01:34.08 | starseeker | has to head out, will try new regex stuff tomorrow |
| 01:34.27 | CIA-14 | BRL-CAD: 03starseeker * r43829 10/brlcad/branches/cmake/src/libged/red.c: MFC r43828 |
| 01:34.49 | CIA-14 | BRL-CAD: 03brlcad * r43830 10/brlcad/trunk/src/libged/red.c: |
| 01:34.50 | CIA-14 | BRL-CAD: simplify with the [:space:] character class which already includes newline and |
| 01:34.50 | CIA-14 | BRL-CAD: carriage return. [:blank:] is a gnu extension and won't necessarily be |
| 01:34.50 | CIA-14 | BRL-CAD: supported for a given vendor regex library. also fix a missed character class |
| 01:34.50 | CIA-14 | BRL-CAD: '?' inclusion with the [:space:] swappage. |
| 01:38.45 | starseeker | sighs - someday I'll understand regex... |
| 01:38.57 | CIA-14 | BRL-CAD: 03starseeker * r43831 10/brlcad/branches/cmake/src/libged/red.c: MFC r43830 |
| 01:42.29 | starseeker | huh - apparently Kitware has a free CDash service: http://my.cdash.org/ |
| 01:43.39 | starseeker | wow http://my.cdash.org/index.php?project=OpenStudio |
| 01:43.41 | brlcad | yep |
| 01:43.49 | CIA-14 | BRL-CAD: 03brlcad * r43832 10/brlcad/trunk/src/libged/red.c: semper fi, er, simplify. let things fall through so there is only one copy of the cleanup code and a consolidated simplified return location. |
| 01:44.24 | starseeker | brlcad: would that serve for a BRL-CAD reporting setup or would we want our own CDash server? |
| 01:44.26 | brlcad | man that looks like ass |
| 01:44.43 | brlcad | I *want* to love cdash .. but that ui is a bit of an eyesore :) |
| 01:44.51 | starseeker | hehe |
| 01:45.06 | brlcad | I'd think we'd still want our own, but still up to whomever does the work in my book |
| 01:45.28 | starseeker | well, we could always rip off the CruiseControl generated html and teach CDash to output that instead... |
| 01:45.28 | brlcad | that way we could hook scripting into the loop for custom actions |
| 01:46.00 | brlcad | it is just the reporting front-end, so you still have to have compilation nodes set up and sending in results |
| 01:46.12 | starseeker | right |
| 01:46.18 | brlcad | cruisecontrol is the other one I couldn't remember |
| 01:46.23 | brlcad | buildbot the third |
| 01:46.36 | starseeker | cruisecontrol had the GUI you liked? |
| 01:47.06 | brlcad | http://cruisecontrol.sourceforge.net/ |
| 01:47.18 | brlcad | it's really easy to use |
| 01:47.55 | brlcad | features neatly grouped together with simplified reporting, then options to dive into more detail or even raw build logs only when you ask for it |
| 01:47.57 | CIA-14 | BRL-CAD: 03starseeker * r43833 10/brlcad/branches/cmake/src/libged/red.c: MFC r43832 |
| 01:48.19 | starseeker | hmm... guess we could use their exec builder to fire off cmake |
| 01:48.49 | CIA-14 | BRL-CAD: 03brlcad * r43834 10/brlcad/trunk/src/libged/red.c: ws cleanup |
| 01:49.38 | brlcad | yep |
| 01:49.49 | brlcad | though again, strong emphasis on "no strong preference" |
| 01:50.20 | brlcad | if you or anyone else wanted to even just play with a small random pet-project CI system, I'd take no issue |
| 01:50.40 | starseeker | is wary of a tool brlcad doesn't like the looks of (*memories of font discussions and graphical layout issues*) |
| 01:50.53 | brlcad | it's much more important that we have continuous integration than we have "the best" |
| 01:51.21 | brlcad | I'd get used to any dashboard so long as it wasn't a maintenance burden or timesink |
| 01:51.50 | starseeker | wonder where we can run enough virtual machines to hit all the configs and not create a pool of molten metal... |
| 01:51.59 | brlcad | their whole purpose is to report failures simply and *quickly* so they get caught as early as possible, so they can get isolated and fixed as early/easily as possible |
| 01:52.06 | starseeker | pity sourceforge took down that server farm |
| 01:52.45 | brlcad | any desktop machine with a few corew would be sufficient enough to crank out a half-dozen builds an hour |
| 01:53.24 | starseeker | dynamic lib only and no docs, maybe... |
| 01:54.07 | starseeker | unless maybe virtual machine overhead is lower these days than I remember |
| 01:55.04 | brlcad | anything that can presently compile the package in less than 10 minutes should be sufficient |
| 01:55.37 | starseeker | wonder if we could prepare stripped down virtual box/qemu images set up to build BRL-CAD and report, then set them up as downloadable goodies for people who want to pile in and and automatically test... setiathome for BRL-CAD :-P |
| 01:56.29 | brlcad | at worse, you get another box or two or have different types of test builds -- quick fast simple reduced builds across everything on one box, then a more extensive slower build box that runs the full regression, docs, bench, etc at a slower rate |
| 01:57.00 | brlcad | heh, I'm not sure that'd be useful :) |
| 01:57.16 | brlcad | if it was a vm image, we'd get tons of reports testing the exact same thing |
| 01:57.35 | starseeker | well, we wouldn't need hundereds of testers - but we could have a list of "images we need testing and need volunteers for" |
| 01:58.06 | brlcad | better would be a build target or even a shell script that took a given checkout or source tarball through all the paces, and reported back |
| 01:58.35 | brlcad | so we could basically get a report anytime someone tried to build |
| 01:58.48 | starseeker | that'd be configuration explosion though |
| 01:59.02 | brlcad | sure, so? :) |
| 01:59.07 | brlcad | entries in a db is all |
| 01:59.17 | brlcad | we hit up the most common failure reports |
| 01:59.28 | brlcad | till there are none |
| 01:59.32 | starseeker | that sounds a bit different than cruisecontrol/cdash... |
| 01:59.44 | brlcad | they'd report to cc/cd |
| 02:00.04 | starseeker | I got the impression cc/cd didn't know the details about the machine/os/etc... |
| 02:00.12 | starseeker | maybe I didn't look close enough |
| 02:00.48 | brlcad | doesn't really have to know, it just has to send in results -- we could easily put machine details into a log that is sent back |
| 02:00.56 | brlcad | benchmark suite does exactly that now |
| 02:01.05 | starseeker | nods |
| 02:01.18 | brlcad | speaking of such |
| 02:01.40 | brlcad | we're going to have to really kick in gear things into gear thursday if gsoc app is going to happen |
| 02:01.58 | starseeker | ah, we're coming up on the deadline? OK |
| 02:02.10 | starseeker | hunts for the wiki page with project ideas |
| 02:02.16 | brlcad | deadline is friday mid-day |
| 02:02.35 | brlcad | it'll take me a couple hours to write up the application submission |
| 02:02.50 | starseeker | k - what do you need me to do firstist and mostist? |
| 02:02.57 | brlcad | few hours for someone to work up a (VERY) detailed project ideas page |
| 02:03.43 | brlcad | they gave feedback in '09 that our ideas page needed a lot more ideas and detail (but that they were going to accept us anyways) |
| 02:03.48 | starseeker | update http://brlcad.org/wiki/Google_Summer_of_Code/Project_Ideas or start over? |
| 02:04.07 | starseeker | s/update/expand upon |
| 02:04.18 | brlcad | I'd start with http://brlcad.org/wiki/Google_Summer_of_Code/, update any references to dates/times/requirements |
| 02:04.31 | brlcad | add new section for 2011 where there are dated sections |
| 02:04.46 | starseeker | right |
| 02:05.22 | brlcad | the project ideas page for 2008 was moved to http://brlcad.org/wiki/Google_Summer_of_Code/2008/Project_Ideas |
| 02:05.43 | brlcad | so can do the same for 2009 and either start a new http://brlcad.org/wiki/Google_Summer_of_Code/Project_Ideas or expand on it |
| 02:06.30 | brlcad | I think we should strongly de-emphasize projects that let devs code alone like we've had in the past |
| 02:06.30 | starseeker | they wanted MORE ideas?? we've got quite a few here |
| 02:06.38 | brlcad | yeah, ironic, eh? |
| 02:07.17 | starseeker | so, no going off in the corner and working on iges or step? |
| 02:07.33 | brlcad | we get compared against other big-boys, they expect more |
| 02:07.36 | brlcad | e.g., http://www.freebsd.org/projects/summerofcode.html#ideas |
| 02:07.54 | brlcad | freebsd's whole set of gsoc pages is arranged decently |
| 02:08.06 | brlcad | notice how those idea links expand to something even more impressive |
| 02:09.16 | brlcad | yeah, ideally not things like iges/step unless they have some exceptional familiarity in that area |
| 02:09.21 | starseeker | brlcad: they can code almost anything alone if they put their minds to it... |
| 02:09.38 | brlcad | something that will make them explore the code better, reuse, refactor, integrate |
| 02:10.06 | brlcad | the only project that might be an exception would be continuation of new gui |
| 02:11.06 | starseeker | nods... I'll do my best... |
| 02:11.31 | starseeker | would continuation of the constraints work fall under the integrate category or the isolated category? |
| 02:12.54 | starseeker | crud - I've gotta get outta here |
| 02:13.30 | brlcad | starseeker: better example of an "integrated" projects that would benefit us you can add/expand: ogre display manager, qt display manager, qt framebuffer, cross-platform adrt, libged transactions, .. maybe a quick scan down TODO for the bigger tasks |
| 02:13.45 | brlcad | constraint continuation would be integrated |
| 02:14.14 | brlcad | goal would have to be something precise like prep or parameters hooked into mged menus, etc |
| 02:15.44 | brlcad | also need mentors to step up: dloman indianlarry ``Erik louipc yukonbob poolio or anyone else |
| 02:15.59 | brlcad | since I'll need your google id again |
| 03:22.13 | starseeker | brlcad: oh - any opinion on using %s for both strings and vls in bu_log? |
| 03:28.30 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r0 10/wiki/Special:Log/move: [[Google Summer of Code/Project Ideas]] moved to [[Google Summer of Code/2009 Project Ideas]]: Moving old Project Ideas page to a date specific location - time for a new one. |
| 03:38.01 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2487 10/wiki/Google_Summer_of_Code/Project_Ideas: Start layout setup for the new Project Ideas iteration |
| 03:49.01 | brlcad | erp, missing slash |
| 03:51.00 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r0 10/wiki/Special:Log/delete: |
| 03:51.00 | CIA-14 | BRL-CAD: deleted "[[Google Summer of Code/Project Ideas]]": content was: 'The list of |
| 03:51.00 | CIA-14 | BRL-CAD: possible projects below should serve as a good starting point for new developers |
| 03:51.00 | CIA-14 | BRL-CAD: that would like to get involved in working on BRL-CAD. Th...' (and the only |
| 03:51.00 | CIA-14 | BRL-CAD: contributor was '[[Special:Contributions/Starseeker|Starseeker]]') |
| 03:51.09 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r0 10/wiki/Special:Log/delete: deleted "[[Google Summer of Code/2009/Project Ideas]]": Deleted to make way for move |
| 03:51.11 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r0 10/wiki/Special:Log/move: [[Google Summer of Code/2009 Project Ideas]] moved to [[Google Summer of Code/2009/Project Ideas]]: consistency with 2008 |
| 03:51.42 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2490 10/wiki/Google_Summer_of_Code/2009: refere to old ideas page |
| 03:52.40 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r0 10/wiki/Special:Log/delete: restored "[[Google Summer of Code/Project Ideas]]": 2 revision(s) restored: hrm, looks like mw or I was confused with a bogus url |
| 03:53.30 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2491 10/wiki/Google_Summer_of_Code/2009: historic ref |
| 03:54.04 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2492 10/wiki/Google_Summer_of_Code/2009/Project_Ideas: historic ref |
| 03:56.34 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2493 10/wiki/Google_Summer_of_Code/2009: past tense |
| 04:00.25 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2494 10/wiki/Google_Summer_of_Code: reword for past tense |
| 04:04.26 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2495 10/wiki/Google_Summer_of_Code: reword for more awesome |
| 04:05.57 | brlcad | starseeker: my inclination is that would be bad (overloaded behavior, inconsistent, probably more confusing than warnings) and probably wouldn't even work since the compiler is supposed to validate that typeof() is a char* for %s, which it wouldn't be |
| 04:10.17 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2496 10/wiki/Google_Summer_of_Code/2011: stub in for 2011 |
| 04:13.37 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2497 10/wiki/Google_Summer_of_Code/Checklist: numbered list |
| 04:14.17 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2498 10/wiki/Google_Summer_of_Code/Checklist: retitle |
| 04:15.28 | starseeker | growl... so we're permenantly stuck with the %V warnings then |
| 04:16.23 | starseeker | unless we get the gcc devs to add a new feature, and even then it'll only be in the newest gcc versions |
| 04:16.53 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2499 10/wiki/Google_Summer_of_Code/Checklist: regroup for clarity |
| 04:17.44 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2500 10/wiki/Google_Summer_of_Code/Checklist: subtasks don't need to be numbered |
| 04:21.57 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2502 10/wiki/Google_Summer_of_Code: KISS |
| 04:23.14 | brlcad | wonders if dloman would be willing to work up this year's gsoc flyer |
| 04:24.04 | brlcad | ( prior examples at http://brlcad.org/wiki/Google_Summer_of_Code#BRL-CAD_participation_in_GSoC ) |
| 04:31.58 | CIA-14 | BRL-CAD: 03Sean 07http://brlcad.org * r2503 10/wiki/Google_Summer_of_Code/Checklist: add deadline dates |
| 04:33.11 | brlcad | that's about all I can push through for tonight |
| 04:33.18 | brlcad | need to get back to release |
| 04:33.20 | starseeker | nods |
| 04:35.10 | brlcad | thinks we need at least 30 fleshed out ideas, perhaps categorized similar to http://brlcad.org/wiki/Contributor_Quickies |
| 04:35.29 | brlcad | (EASY, MEDIUM, HARD) |
| 04:36.06 | brlcad | or Web, C, C++, Tcl |
| 04:36.15 | brlcad | both |
| 04:36.36 | starseeker | brlcad: OK... I'm just tossing stuff up right now, I'll sort it once it's more fleshed out |
| 04:36.42 | brlcad | nods |
| 04:37.11 | brlcad | you can probably utilize the entire 2009 list and just keep adding more, remove the one or two that are no longer relevant |
| 04:38.03 | starseeker | oh - I was starting with the whole "more interactive" thing and was gonna review the 2009 list to see what qualified... |
| 04:39.26 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2504 10/wiki/Google_Summer_of_Code/Project_Ideas: More ideas - this isn't anything like final form, probably - just scratchpad |
| 04:40.30 | brlcad | typo in the top title section |
| 04:40.45 | brlcad | sure, however you wanna do it :) |
| 04:40.58 | starseeker | the "AN IDEA" one? |
| 04:40.59 | brlcad | aim for 50, hopefully we'll get to 40 :) |
| 04:41.05 | brlcad | "=." |
| 04:41.17 | starseeker | oh, good eye |
| 04:41.41 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2505 10/wiki/Google_Summer_of_Code/Project_Ideas: |
| 04:41.47 | starseeker | kinda liked FreeBSD's grouping by larger scale topic |
| 04:42.09 | starseeker | I'll take a whack, then you can tell me why it's wrong :-P |
| 04:42.26 | starseeker | brlcad: did you want to get those red changes into this release? |
| 04:45.59 | *** join/#brlcad Ralith (~ralith@S010600221561996a.vc.shawcable.net) |
| 04:56.00 | *** join/#brlcad dli (~dli@dsl-69-171-148-245.acanac.net) |
| 04:57.26 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2506 10/wiki/Google_Summer_of_Code/Project_Ideas: list a few more things... |
| 04:58.00 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2507 10/wiki/Google_Summer_of_Code/Project_Ideas: Constraint solver needs its own page |
| 04:58.33 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2508 10/wiki/Geometric_Constraint_Solver: Put description on individual page |
| 05:09.28 | starseeker | brlcad: do you happen to know whether or not its possible to recognize an arbitrary NURBS surface as being a component of (say) a sphere or a torus? |
| 05:10.30 | Ralith | starseeker: is BRL-CAD doing SoC this year? |
| 05:10.35 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2509 10/wiki/Google_Summer_of_Code/Project_Ideas: More tasks to flesh out... |
| 05:10.43 | starseeker | we'll see :-) |
| 05:11.24 | Ralith | if so, will almost certainly submit a GPU accel proposal. |
| 05:18.59 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2510 10/wiki/Google_Summer_of_Code/Project_Ideas: Couple more NURBS ideas... |
| 05:26.26 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2511 10/wiki/Ayam_Editor_Feature_Integration: Toss a few notes into the Ayam/NURBS entry |
| 05:53.25 | brlcad | starseeker: did someone report red? |
| 05:53.31 | starseeker | Victor |
| 06:06.27 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2512 10/wiki/Google_Summer_of_Code/Project_Ideas: That's around 30 ideas... some work to do tomorrow |
| 06:50.18 | brlcad | Ralith: one never knows until orgs are announced |
| 06:51.02 | Ralith | yeah, meant to ask "applying" |
| 06:51.07 | Ralith | and wiki confirms |
| 06:51.44 | brlcad | starseeker: not definitively unless the surfaces are tagged with originating geometry |
| 06:53.02 | brlcad | Ralith: yeah, it looks like we're probably applying if I can get a couple more mentors to sign on |
| 06:53.10 | Ralith | awesome |
| 06:53.32 | Ralith | would be a great way to spend the summer |
| 06:54.36 | brlcad | Ralith: I'd be cautious about a gpu accel project for the same aforementioned "prefer integrated code, minimal 'new code'" concerns -- it'd have to be very specific about how it'd be integrated and beneficial |
| 06:55.35 | brlcad | and from a former gsoc'er, you'll have to work twice as hard to convince |
| 06:56.19 | Ralith | hm, that's a shame |
| 06:57.12 | brlcad | not really, it's only fair to the 3k other students :) |
| 06:57.21 | Ralith | hah |
| 06:57.23 | Ralith | yeah |
| 06:58.23 | brlcad | if someone had been more actively commiting over the past year and a half, it'd be a totally different story ;) |
| 06:58.24 | Ralith | may play with it anyway, depending on what else I end up doing |
| 06:58.27 | brlcad | hell, even passively :) |
| 06:59.29 | Ralith | looks slightly regretful |
| 07:00.08 | brlcad | the point of gsoc is to get students engaged in open source, even if it's not brl-cad ... http://www.ohloh.net/p/brlcad/contributors/17162689320215 tells a sad story |
| 07:00.45 | Ralith | yes, it's quite true |
| 07:00.51 | Ralith | though I've been plenty active elsewhere |
| 07:00.56 | Ralith | not sure why ohloh hasn't picked up on that |
| 07:01.10 | brlcad | not intending to pick on you personally, just saying that puts you back into the boat with everyone else on level standing |
| 07:01.18 | Ralith | I know |
| 07:01.30 | brlcad | you can combine accounts |
| 07:02.07 | Ralith | I think most of the projects simply aren't on ohloh |
| 07:02.11 | Ralith | which can be remedied, certainly |
| 07:02.37 | Ralith | but anyway, not too relevant |
| 07:03.24 | brlcad | yeah, the project itself is more the issue, it'd have to be carefully spelled out |
| 07:03.50 | Ralith | nod |
| 07:04.19 | brlcad | for example, most of the gpgpu research is on accellerating triangle raytracing -- that's just not very interesting any more, especially now that we have our new TIE library integrated with LIBRT |
| 07:04.23 | Ralith | mainly it strikes me as a good opportunity to learn more about raytracing and rt's internals in particular, as well as GPU frobbery |
| 07:04.52 | Ralith | hm. |
| 07:05.03 | Ralith | I wasn't aware it was a research area per se |
| 07:05.04 | brlcad | so it'd have to be some other means to accellerate librt, which would be either in the spatial partitioning, the boolean weaving, the primitive evaluations, etc |
| 07:05.14 | brlcad | oh hell yeah, a huge one |
| 07:05.40 | Ralith | isn't raytracing embarassingly parallel across each ray? |
| 07:05.55 | brlcad | nvidia and intel fighting it out, both trying to push gpgpu raytracing into gaming, it's been five years+ in the making |
| 07:06.02 | brlcad | yes? |
| 07:07.06 | Ralith | perhaps I drastically misunderstand the problem, but wouldn't simply mapping rays across GPU compute unitsâwith fairly straight port of the relevant rt codeâbe a huge gain? |
| 07:07.39 | brlcad | mapping rays to the gpu is trivial, days work |
| 07:07.52 | brlcad | you have to evaluate and march those rays |
| 07:07.57 | brlcad | against geometry |
| 07:08.06 | Ralith | right, I meant mapping the entire process |
| 07:08.48 | brlcad | so how does one evaluate whether a ray intersects with a torus on the gpu? |
| 07:09.00 | brlcad | heck, even a simple sphere |
| 07:09.27 | brlcad | it's doable, but it's not a 1-1 mapping or a mere matter of "porting" anything |
| 07:09.49 | brlcad | you generally have to completely rethink how data is packed and evaluated |
| 07:10.46 | Ralith | I guess it should have been obvious that the hardware is ill-optimized for the subtleties of geometry more complex than triangles. |
| 07:10.49 | brlcad | that's why most of the research is just with triangles -- only have to deal with one entity type that is very trivial to evaluate, has a fixed constant data size, etc |
| 07:11.10 | Ralith | that sounds a bit beyond what I can accomplish based on what I know, at least in a single summer. |
| 07:11.22 | brlcad | the hardware isn't even really geared for triangles, but the calculations are so few |
| 07:11.31 | Ralith | well |
| 07:11.37 | Ralith | sketchy understanding here |
| 07:12.02 | brlcad | you actually *can* do spheres even easier if you apply the same techniques you apply for triangles, but then a beast like a torus becomes super problematic because it requires a root solver |
| 07:12.03 | Ralith | but aren't GPUs pretty much all about vector ops, and triangles trivial to work with within those? |
| 07:12.21 | brlcad | yep, that's where fixed data size becomes a win |
| 07:12.30 | brlcad | *small* fixed data size |
| 07:12.43 | Ralith | and the math, surely? |
| 07:12.58 | Ralith | well, I suppose you just said that |
| 07:13.12 | Ralith | alright, thanks for the explanation; sounds like this was ill-conceived. |
| 07:14.03 | Ralith | an interesting problem to be sure, but I don't have the background to do original work in that sort of thing. |
| 07:14.45 | brlcad | it's a great feature, a great research project, but very non-trivial to say the least -- you'd probably be able to publish in an journal or even siggraph if you got it working for general implicit geometry |
| 07:15.21 | brlcad | easier attack vector would be accelerating our boolean weaving step |
| 07:15.30 | Ralith | haven't a clue what that is O.o |
| 07:15.36 | brlcad | exactly :) |
| 07:16.19 | Ralith | ah well. |
| 07:16.23 | brlcad | that's where the "you'd have to work twice as hard to convince" came from |
| 07:16.48 | brlcad | because it could take a month just to get schooled up on the background knowledge needed |
| 07:17.08 | Ralith | knowing the scope of the problem, even that seems generous |
| 07:17.18 | Ralith | for weak values of knowing even now |
| 07:18.06 | brlcad | that's where if you were already familiar with even the basic mechanics of librt working on portions of the code consistently, that'd make for an easier case to dive into such a hard problem |
| 07:19.30 | brlcad | you'd know the application structure's purpose, how hit callbacks work, when boolean evaluation is performed, what that means, how the implicits evaluate, types of CSG dag nodes, etc |
| 07:20.22 | Ralith | perhaps continuing jonored's work on a toolpather would be more realistic? |
| 07:20.38 | brlcad | fairly complex code, highly optimized, and critical code at that which would require 100% validation |
| 07:20.49 | Ralith | though I'm not certain of my ability to find a footing with the (as far as I can tell, currently buggy) math there either |
| 07:21.11 | brlcad | what toolpather? |
| 07:21.20 | Ralith | I suppose he never posted it here |
| 07:21.31 | Ralith | you recall a while back that he was working on a gcode generator? |
| 07:21.34 | Ralith | couple years ago or so? |
| 07:21.45 | brlcad | not specifically |
| 07:21.50 | *** join/#brlcad cjdevlin (~devlin@d118-75-252-178.try.wideopenwest.com) |
| 07:21.58 | Ralith | he was, and it got abandoned in favor of his studies |
| 07:22.09 | Ralith | but I got ahold of him a few months back and got him to send me a tarball of the git repo |
| 07:22.28 | brlcad | I do recall that discussion |
| 07:22.44 | Ralith | he claims outline tracing is mostly working, though in my limited experiments I've not gotten correct output |
| 07:22.51 | brlcad | that wasn't based on brl-cad iirc, though, no? |
| 07:23.09 | Ralith | er, it uses librt to interrogate BRL-CAD databases |
| 07:23.21 | brlcad | hrm |
| 07:23.22 | Ralith | not sure how much more 'based on' it gets |
| 07:23.36 | Ralith | want a look at the tarball? |
| 07:23.55 | brlcad | not at the moment, but it is interesting |
| 07:23.59 | Ralith | kk |
| 07:24.12 | Ralith | I'll see if I can take a better look at it in the coming month and work out what state it's really in |
| 07:26.20 | brlcad | yeah, that was back in january we were talking about it |
| 07:26.28 | brlcad | you were asking about curvature |
| 07:28.31 | Ralith | yeah, jonored mentioned that torus curvature was the standing bug; I had planned to dive in and fix that if I could get it to behave for simpler primitives |
| 07:28.52 | Ralith | haven't managed that yet, though haven't put much time in it |
| 07:37.13 | brlcad | if it really is non-gui based (which an old post of his indicated), that would be a good one to get working and fully integrate |
| 07:37.25 | brlcad | basically a "g-gcode" converter |
| 07:45.21 | Ralith | it is indeed commandline. |
| 07:46.00 | Ralith | in fact it even follows the naming scheme |
| 07:46.01 | Ralith | g2gcode |
| 07:46.08 | Ralith | bear in mind, though, that additive and subtractive toolpaths are differnet beasts |
| 07:46.16 | Ralith | different* |
| 07:57.27 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) |
| 07:58.53 | Ralith | gah. |
| 07:59.14 | Ralith | journal website lists this article as 'full text available,' but when I log in through my univ it's preview only :| |
| 07:59.42 | Ralith | gunna have to get a physical copy I guess. |
| 08:08.57 | Ralith | (trying to get ahold of Martin Held's "On the computational geometry of pocket machining") |
| 08:21.55 | *** join/#brlcad Stattrav (~Stattrav@122.167.172.122) |
| 08:21.56 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 10:38.38 | *** join/#brlcad dli (~dli@dsl-69-171-148-245.acanac.net) |
| 11:16.19 | d_rossberg | i get fringes on ray-tracing cones |
| 11:17.26 | d_rossberg | e.g. the havoc's tail (ae 300 0) |
| 13:47.57 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2513 10/wiki/Google_Summer_of_Code/Project_Ideas: add subversion idea |
| 13:55.10 | starseeker | brlcad: would search exec be a viable GSoC project? |
| 14:45.37 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2514 10/wiki/Google_Summer_of_Code/Project_Ideas: /* BRL-CAD Core Library Enhancement */ |
| 14:57.37 | ``Erik | cdash is heavily css driven. if it's the look and not the structure, it should be easy to make it less ugly |
| 15:20.32 | CIA-14 | BRL-CAD: 03erikgreenwald * r43835 10/geomcore/trunk/tests/utility/threadTest.cxx: programmatically test for functional threading instead of spewing junk to the screen for human analysis |
| 15:21.46 | starseeker | ``Erik: you good to mentor? |
| 15:23.06 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2515 10/wiki/Google_Summer_of_Code/Project_Ideas: Couple more line items - that's 38... |
| 15:23.19 | ``Erik | yup, msg'd brlcad my id |
| 15:24.34 | starseeker | ``Erik: can you scan that list and see if I left out any juicy items? brlcad was talking about wanting 50... |
| 15:24.55 | ``Erik | yeah, I looked over it t his morning, um |
| 15:25.16 | *** join/#brlcad juanman (~quassel@unaffiliated/juanman) |
| 15:25.17 | ``Erik | I'll splat some stuff on and if someone doesn't like it, they can remove it O.o |
| 15:25.25 | *** join/#brlcad Stattrav (~Stattrav@117.192.159.181) |
| 15:25.25 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 15:31.44 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2516 10/wiki/Google_Summer_of_Code/Project_Ideas: /* BRL-CAD Core Library Enhancement */ |
| 15:32.33 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2517 10/wiki/Google_Summer_of_Code/Project_Ideas: /* Tools */ |
| 15:36.03 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2518 10/wiki/Google_Summer_of_Code/Project_Ideas: /* Tools */ |
| 15:36.47 | ``Erik | is [[NURBS Intersections]] an evaluation of a CSG of NURBS into a single one? |
| 15:37.10 | starseeker | no, that's just the surface/surface problem and related subproblems |
| 15:37.40 | starseeker | I doubt a summer of code student could get all the way to the final evaluated model |
| 15:37.57 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2519 10/wiki/Google_Summer_of_Code/Project_Ideas: 'csg' makes more sense than 'boolean' here |
| 15:38.58 | CIA-14 | BRL-CAD: 03Erik 07http://brlcad.org * r2520 10/wiki/Google_Summer_of_Code/Project_Ideas: /* BRL-CAD Core Library Enhancement */ |
| 15:39.25 | ``Erik | hm, can always harvest TODO and http://brlcad.org/~sean/ideas.html for some more |
| 15:39.40 | starseeker | nods - I grabbed a few out of there |
| 15:39.49 | starseeker | was trying to look for things that would be 'integrated' |
| 15:39.53 | ``Erik | orange page, too? |
| 15:39.59 | starseeker | nods |
| 15:40.20 | starseeker | for example, I'd worry that a Blender plugin would kinda be off in its own little world... |
| 15:41.01 | starseeker | now if they want to snarf the Blender file import that game kit implemented and turn it into a BRL-CAD importer... :-) |
| 15:41.24 | ``Erik | a little bit, but it would need a fair amount of communication to figure out how to translate and what functionality |
| 15:42.06 | starseeker | nods - well, we've probably got close to enough lined up there - now we need to start filling 'em in |
| 15:42.11 | ``Erik | ooh, they have one now? last I looked, their format was basically a swizzle of memory dumped to disk |
| 15:42.37 | starseeker | yeah, pretty much - apparently someone figured out how to do something intelligent with it... |
| 15:42.41 | starseeker | one sec... |
| 15:42.55 | ``Erik | starts writing a test for ControlledThread O.o |
| 15:43.06 | starseeker | http://code.google.com/p/gamekit/ |
| 15:45.15 | ``Erik | yeah, saw that before, hm |
| 15:48.18 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2521 10/wiki/Google_Summer_of_Code/Project_Ideas: |
| 16:21.27 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2522 10/wiki/OGRE_Display_Manager: Flesh out OGRE task (I think this is low difficulty?) |
| 16:21.40 | *** join/#brlcad piksi (piksi@pi-xi.net) |
| 16:45.24 | brlcad | starseeker: sure, search exec would be viable |
| 16:47.28 | brlcad | most of the index card tasks are decent gsoc tasks |
| 16:47.33 | brlcad | should do a quick scan |
| 16:48.12 | brlcad | since they're < month for us, that easily scales up to 3 months for someone else including communication, reporting, testing, and documentation |
| 17:11.35 | dli | brlcad, http://paste.pocoo.org/show/351409/ |
| 18:36.43 | CIA-14 | BRL-CAD: 03erikgreenwald * r43836 10/geomcore/trunk/tests/utility/threadTest.cxx: start stubbing in ControlledThread test |
| 18:41.51 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2523 10/wiki/Qt_Display_Manager: Qt display manager |
| 18:42.31 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2524 10/wiki/Qt_Display_Manager: Qt, not ogre |
| 18:53.59 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2525 10/wiki/Qt_Framebuffer: Qt framebuffer |
| 18:56.07 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2526 10/wiki/Qt_Framebuffer: Grr - framebuffer, not display manager |
| 19:08.51 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2527 10/wiki/Other_Cross_Platform_Framebuffer: Cross Platform framebuffer |
| 19:21.20 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2528 10/wiki/High_Dynamic_Range_Support: HDR output support |
| 19:21.27 | starseeker | ``Erik: mind reviewing the HDR one if you have a sec? |
| 19:21.57 | ``Erik | sure, um, and I just changed the css to mark the 'new' class as red, so'z you can tell which ones you've done |
| 19:22.11 | starseeker | thanks :-) |
| 19:22.30 | ``Erik | (we can revert it once this exercise is over if brlcad wants, it's in RCS) |
| 19:23.12 | ``Erik | "There is not fundamental reason" ... wow, right out of the gate :> |
| 19:23.29 | starseeker | heh, sorry - distracted |
| 19:23.35 | starseeker | (house is leaking again) |
| 19:23.50 | ``Erik | other than the picasso grammar, looks decent... |
| 19:28.10 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2529 10/wiki/High_Dynamic_Range_Support: |
| 19:40.18 | CIA-14 | BRL-CAD: 03Starseeker 07http://brlcad.org * r2530 10/wiki/MGED_to_Archer_Command_Migration: Command migration |
| 20:09.10 | CIA-14 | BRL-CAD: 03erikgreenwald * r43837 10/geomcore/trunk/tests/GS/GeometryServiceTest.cxx: indent |
| 20:23.33 | CIA-14 | BRL-CAD: 03erikgreenwald * r43838 10/geomcore/trunk/tests/GS/GeometryServiceTest.cxx: purdee kulurz |
| 21:01.24 | CIA-14 | BRL-CAD: 03Markhobley 07http://brlcad.org * r2531 10/wiki/Talk:Error_Messages: wikibook |
| 21:27.22 | brlcad | dli: oh snap |
| 21:27.28 | brlcad | aua he left |
| 21:31.19 | brlcad | debates trying an application form this year |
| 21:31.34 | starseeker | for GSoc you mean? |
| 21:31.38 | brlcad | yeah |
| 21:31.49 | starseeker | does seem kinda rushed |
| 21:32.14 | brlcad | many of them just have no idea on how to pitch an idea, so there ends up being a lot of information we have to keep asking for |
| 21:33.35 | brlcad | it has helped in the past to separate out those who HAVE thought through their idea and have a lot to say, but many of those devs tend to have trouble integrating their thoughts with our codebase |
| 21:33.50 | brlcad | or worse, the ones that can write great, but can't actually code |
| 21:34.45 | starseeker | had thought about some basic "submit this code with your application" kinda things, but that takes time to define |
| 21:35.22 | CIA-14 | BRL-CAD: 03erikgreenwald * r43839 10/geomcore/trunk/tests/GS/GeometryServiceTest.cxx: more indentation, staticizing, etc |
| 21:35.26 | starseeker | ``Erik: what's your take? |
| 21:35.55 | ``Erik | *shrug* |
| 21:36.33 | ``Erik | we tried "submit a patch" a couple years back, seems it was a bit of a pain |
| 21:37.15 | starseeker | ``Erik: should we rush to get it "ready" or punt until next year? |
| 21:37.47 | brlcad | submit a patch worked pretty well I thought -- the strong patches ended up being the strong coders |
| 21:38.09 | brlcad | kept the application count down, but I think it worked |
| 21:38.23 | ``Erik | the weak patches were a depressing time sink, though :) |
| 21:38.41 | brlcad | so we be more agressive in reviewing them |
| 21:39.03 | ``Erik | starseeker: doesn't hurt to try, they're usually fairly easy going if minor tweaking is needed |
| 21:39.04 | brlcad | we can point them to the quickies this year |
| 21:39.22 | starseeker | brlcad: good idea |
| 21:39.26 | brlcad | no quickie should take more than a day or few |
| 21:39.35 | starseeker | whoops, contractor is here |
| 21:39.37 | brlcad | we can certainly expand that list |
| 21:57.01 | CIA-14 | BRL-CAD: 03erikgreenwald * r43840 10/geomcore/trunk/ (59 files in 3 dirs): footer.sh |
| 22:00.37 | ``Erik | damnit, thought I broke that soon enough |
| 22:33.17 | starseeker | brlcad: expand the projects list? I've still got to get all those fleshed out |
| 22:47.33 | *** join/#brlcad dli (~dli@dyn-216-212.wireless.concordia.ca) |
| 23:56.42 | brlcad | starseeker: I meant the quickies list, no worries -- that'd be before they start applying, not before tomorrow |
