| 00:13.24 | CIA-77 | BRL-CAD: 03starseeker * r43530 10/brlcad/branches/cmake/CMakeLists.txt: Remove the complex and only partially successful noprod logic - with targets in toplevel bin dirs anyway the utility is minimal, and not worth the complexity. |
| 00:36.45 | CIA-77 | BRL-CAD: 03starseeker * r43531 10/brlcad/branches/cmake/ (3 files in 3 dirs): Ignore other in src |
| 00:39.42 | CIA-77 | BRL-CAD: 03starseeker * r43532 10/brlcad/branches/cmake/src/librt/CMakeLists.txt: Ignore search.h |
| 00:50.20 | *** join/#brlcad Klebel (~mk@w73.RIC.Berkeley.EDU) |
| 00:50.49 | Klebel | I can't find 'Set H' in the Edit menu |
| 00:57.36 | Klebel | press "Set H" - says unknown operation. |
| 00:57.45 | Klebel | on the command line |
| 00:58.31 | starseeker | Klebel: we need more context |
| 00:59.29 | Klebel | page 60 in the Introduction to MGED manual. |
| 00:59.29 | Klebel | pdf |
| 01:00.22 | Klebel | I created a right circular cylinder, then it tells me to do, "Edit and then Set H" |
| 01:00.33 | Klebel | and click the middle mouse button several times |
| 01:01.02 | starseeker | to edit a primitive, you need to use sed |
| 01:01.08 | starseeker | that puts you in edit mode |
| 01:02.05 | starseeker | if you aren't seeing Set H you probably aren't in edit mode |
| 01:02.08 | Klebel | ah, so sed base1.s |
| 01:02.08 | Klebel | thanks, Set H is now in the menu, and works |
| 01:02.29 | starseeker | that tutorial was created with an older version of BRL-CAD, so there are occasional differences |
| 01:02.43 | Klebel | yea I've noticed that :/ |
| 01:07.15 | Klebel | is there a command to get out of edit mode? |
| 01:07.26 | starseeker | accept or reject |
| 01:07.40 | Klebel | ok thanks |
| 01:09.02 | CIA-77 | BRL-CAD: 03starseeker * r43533 10/brlcad/branches/cmake/ (27 files in 22 dirs): MFC r43532 |
| 01:10.42 | CIA-77 | BRL-CAD: 03starseeker * r43534 10/brlcad/branches/cmake/src/other/CMakeLists.txt: Uncomment -w again for src/other |
| 01:17.39 | CIA-77 | BRL-CAD: 03starseeker * r43535 10/brlcad/branches/cmake/misc/CMake/CompilerFlags.cmake: We're going for gnu99 now |
| 02:01.02 | CIA-77 | BRL-CAD: 03brlcad * r43536 10/brlcad/trunk/src/bwish/cadAppInit.c: include bin.h instead of winsock2.h |
| 03:28.05 | *** join/#brlcad guest_tttt (~rm@123.136.11.66) |
| 03:34.43 | *** part/#brlcad guest_tttt (~rm@123.136.11.66) |
| 04:32.23 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 04:33.44 | CIA-77 | BRL-CAD: 03brlcad * r43537 10/brlcad/trunk/src/rttherm/pixtest.c: check fwrite return value |
| 04:36.11 | CIA-77 | BRL-CAD: 03brlcad * r43538 10/brlcad/trunk/src/sig/ (24 files): check fwrite return values for failure |
| 04:39.55 | CIA-77 | BRL-CAD: 03brlcad * r43539 10/brlcad/trunk/src/proc-db/brepintersect.cpp: compiler is complaining about the first param to SegmentPolylineIntersect possibly being NULL. as this is dev code, just comment out for now in leu of removing the code. |
| 04:44.14 | CIA-77 | BRL-CAD: 03brlcad * r43540 10/brlcad/trunk/src/mged/points/points_parse.y: bison is being stupid with some output code generating size_t comparisons against >= 0. quell that warning along with a couple other preprocessor symbols that are not defined but being used in expressions. |
| 04:45.52 | CIA-77 | BRL-CAD: 03brlcad * r43541 10/brlcad/trunk/src/ (mged/points/points_scan.l tab/script.l): quell flex lameness where fwrite() is being called without checking the return value. this quiets the compiler. |
| 04:51.08 | CIA-77 | BRL-CAD: 03brlcad * r43542 10/brlcad/trunk/src/ (23 files in 7 dirs): |
| 04:51.09 | CIA-77 | BRL-CAD: categorically check return values for some of the stdio and stdlib routines |
| 04:51.09 | CIA-77 | BRL-CAD: (e.g. fwrite, scanf, system, dup, pipe, ...). not willing to put forth |
| 04:51.09 | CIA-77 | BRL-CAD: time/effort to do anything more than print the error since would have to |
| 04:51.09 | CIA-77 | BRL-CAD: evaluate each call on a case by case basis (and that's not fun). |
| 05:10.50 | *** join/#brlcad dli (~dli@dsl-67-204-45-87.acanac.net) |
| 05:35.23 | *** join/#brlcad Stattrav (~Stattrav@122.167.254.137) |
| 05:35.23 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 06:08.23 | CIA-77 | BRL-CAD: 03brlcad * r43543 10/brlcad/trunk/src/libfb/if_X24.c: fix keybindings on Mac OS X so that cmd-click will produce button 3 events (so we can close framebuffers) without needing the user to set X11.app to emulate a three button mouse. |
| 06:09.28 | CIA-77 | BRL-CAD: 03brlcad * r43544 10/brlcad/trunk/src/libfb/if_ogl.c: do the same mouse-3 binding for ogl |
| 06:51.11 | Klebel | 1 thing I keep noticing in the Introduction manual, is that when I copy primatives through the command line then run sed on the newly copied primative it says: Error sph2.s not being displayed |
| 06:51.37 | Klebel | the only way I am able to copy is through the Primitive Editor |
| 07:01.04 | CIA-77 | BRL-CAD: 03brlcad * r43545 10/brlcad/trunk/src/tclscripts/mged/bindings.tcl: |
| 07:01.04 | CIA-77 | BRL-CAD: fix mged zoom bindings on mac os x with default X11 (where 3-button mouse |
| 07:01.04 | CIA-77 | BRL-CAD: emulation is disabled) so that you can actually zoom out. make cmd-click behave |
| 07:01.04 | CIA-77 | BRL-CAD: the same as button-3. unfortunately, the same binding does not seem possible |
| 07:01.04 | CIA-77 | BRL-CAD: with option-click for mouse 2 (in fact, option seems to act like mod2 aka the |
| 07:01.04 | CIA-77 | BRL-CAD: command-key. none of the other mod types seem to help. |
| 07:07.37 | CIA-77 | BRL-CAD: 03brlcad * r43546 10/brlcad/trunk/NEWS: |
| 07:07.38 | CIA-77 | BRL-CAD: fixed a problem being able to zoom in with the default x11.app configuration |
| 07:07.38 | CIA-77 | BRL-CAD: where you could zoom out, but not back in without enabling 3-button emulation. |
| 07:07.38 | CIA-77 | BRL-CAD: this binds command+button1 the same as button3. couldn't get option+button1 to |
| 07:07.38 | CIA-77 | BRL-CAD: behave as button2 though. |
| 07:10.21 | CIA-77 | BRL-CAD: 03brlcad * r43547 10/brlcad/trunk/src/librt/primitives/table.c: tested and rt_generic_xform() is NOT sufficient. wrong plot and trace with loads of error. |
| 07:10.47 | brlcad | Klebel: when you copy something, it's not automatically drawn |
| 07:10.57 | brlcad | cp old new ; e new |
| 07:11.07 | brlcad | then sed and friends will work |
| 07:11.12 | brlcad | (on new) |
| 07:11.25 | brlcad | e == draw |
| 07:13.30 | brlcad | starseeker: mged regression is failing with "Tcl_Import ERROR: unknown namespace in import pattern "::itcl::*" |
| 07:14.09 | brlcad | wondering if you're seeing that on your end |
| 07:14.50 | brlcad | only namespace or itcl change that comes to mind is the mged/bwish setup routine |
| 07:15.47 | brlcad | yeah, namespace import fails because itcl_init is failing, can't find/load the itcl .so file (testing on linux atm) |
| 07:52.39 | Klebel | thanks brlcad |
| 10:18.24 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
| 10:38.35 | *** join/#brlcad Elrohir (~kvirc@p5B149820.dip.t-dialin.net) |
| 11:07.56 | *** join/#brlcad Elrohir (~kvirc@p5B149820.dip.t-dialin.net) |
| 11:13.53 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 11:30.56 | dloman | Mernin all. |
| 11:46.40 | CIA-77 | BRL-CAD: 03d_rossberg * r43548 10/brlcad/trunk/include/bin.h: |
| 11:46.40 | CIA-77 | BRL-CAD: undef some windows defines which are used in an other context here |
| 11:46.40 | CIA-77 | BRL-CAD: (same as in bio.h) |
| 11:49.56 | CIA-77 | BRL-CAD: 03d_rossberg * r43549 10/brlcad/trunk/ (3 files in 3 dirs): raytrace.h got a regex.h include -- added the corresponding search path |
| 12:14.00 | starseeker | brlcad: I'll have to check - I'm doing most of my development work with CMake these days, so I haven't tried the autotools regression in a while |
| 12:14.12 | starseeker | I thought it worked, but maybe it was a fluke... |
| 12:27.29 | dloman | Huh. Neato: http://sewelldirect.com/Sewell-Minideck-USB-to-DVI-Display-Adapter.asp |
| 12:45.21 | CIA-77 | BRL-CAD: 03starseeker * r43550 10/brlcad/trunk/src/fbserv/fbserv.c: Shadowing a global - fixed |
| 12:47.24 | CIA-77 | BRL-CAD: 03starseeker * r43551 10/brlcad/trunk/src/lgt/ (hmenu.c lgt.c): Remove defined-but-unused functions causing lgt failures. |
| 12:49.29 | starseeker | brlcad: still getting a script.c failure: |
| 12:49.40 | starseeker | script.c: In function âyy_get_next_bufferâ: |
| 12:49.40 | starseeker | script.c:1414: error: comparison between signed and unsigned integer expressions |
| 13:07.12 | starseeker | brlcad: OK, I see a failure |
| 13:07.22 | starseeker | looks like a different one though |
| 13:07.38 | starseeker | brlcad: are you doing an in-dir or out of dir build? |
| 13:09.15 | CIA-77 | BRL-CAD: 03starseeker * r43552 10/brlcad/trunk/src/mged/chgview.c: Don't shadow basename |
| 13:11.16 | starseeker | god I wish we could switch to CMake |
| 13:12.26 | starseeker | out of dir autotools build can't find the tclscripts to initialize the gui, from the looks of it |
| 13:13.49 | starseeker | bizarrely enough, if I start up with nu, package require Itcl DOES succeed |
| 13:15.27 | starseeker | confound it, autopath STILL has /usr/lib at the head of the auto_path search path |
| 13:21.18 | dloman | so, brlcad (autotools) doesn't like out of src builds atm? |
| 13:22.00 | starseeker | oh, I builds OK but it doesn't seem to run |
| 13:23.27 | starseeker | auto_path is getting all the paths set to the build dir versions of tclscript paths, which doesn't work because they're still in the src dir |
| 13:25.57 | starseeker | we can either copy the tclscripts over into the build dir (which is a variation on what CMake does currently) or tweak the path logic to point back to the src dir |
| 13:26.32 | starseeker | but that won't fix the issue of /usr/lib being up front in the search path, which is an excellent way to pull in system installed things and mix the tcl/tk environment up royally |
| 13:30.10 | CIA-77 | BRL-CAD: 03starseeker * r43553 10/brlcad/branches/cmake/ (66 files in 21 dirs): MFC r43552 |
| 13:33.10 | starseeker | nice little subtle issues, particularly when we have to use a local tweaked version of something and there's a vanilla system version getting pulled in instead |
| 13:40.23 | starseeker | saddles up |
| 13:56.42 | ``Erik | <PROTECTED> |
| 14:14.45 | dloman | I love it. |
| 14:15.05 | dloman | the IT guys are trying to tell me that its not possible to drive 4 monitors from two video cards. |
| 14:17.24 | dloman | stares at brlcad's 5 monitor setup and laughs at IT. |
| 14:17.55 | dli | dloman, two video cards for 5 monitor? |
| 14:18.09 | dloman | brlcad actually has 3 cards |
| 14:18.39 | dloman | but the point remains the same. the IT guys I'm emailing obviously dunt know what they are talking about. |
| 14:18.41 | dli | dloman, so X-server can handle as many as you can provide? |
| 14:19.07 | dloman | should |
| 14:22.00 | dli | dloman, nice to know |
| 14:22.51 | dloman | stumbled upon a USB to DVI converter (see link above) that allows up to 6 external monitors :) |
| 14:22.57 | dloman | ....i really want to try that out. |
| 14:24.07 | dli | dloman, I got a thinkpad with faulty video card, maybe, I can drive up a monitor by USB, but I'm not sure |
| 14:24.40 | dloman | Might be worth a look. |
| 14:24.51 | dloman | is the video card integrated or dedicated? |
| 14:25.38 | CIA-77 | BRL-CAD: 03erikgreenwald * r43554 10/brlcad/trunk/src/libfb/if_ogl.c: event is a struct, not a pointer |
| 14:29.08 | dloman | ahahaha, just found a MB that has 7 x16 PCIe slots. Get 7 Eyefinity6's and thats... 42 monitors plus 6 on USB. 48 screens. Awesome. |
| 14:30.05 | dloman | Heh, i wonder if that would actually work :) |
| 14:31.28 | starseeker | consideres the power demand and winces slightly |
| 14:31.57 | dloman | If you can afford all those cards and displays, the Powersupply becomes trivial :) |
| 14:32.09 | ``Erik | mr fusion |
| 14:32.18 | dli | dloman, it's deciated video, but still integrated on the mobo |
| 14:32.40 | ``Erik | 1.21 jiggawatts (whatever a jigga is) |
| 14:32.54 | dloman | dli: suckage. You opened the case to ensure the gfx card isn't removable? |
| 14:33.58 | dloman | ``Erik: you're borderline racist, so be careful :) |
| 14:34.11 | ``Erik | O.o |
| 14:34.12 | dli | dloman, yes, I opened it several times :) right now, the thinkpad works as a file server, and a wireless router for my VoIP |
| 14:34.20 | dloman | ahaha, nice. |
| 14:35.12 | dli | dloman, the VoIP box supports only wired net, so the thinkpad hooks VoIP to wifi. |
| 14:35.20 | dloman | well that usb2dvi thingy costs about 100 bucks, and you might be able to get an entire replacement laptop (depending on how old it is) for that. |
| 14:35.42 | dloman | hates wifi. too slow. |
| 14:36.36 | CIA-77 | BRL-CAD: 03starseeker * r43555 10/brlcad/branches/cmake/src/libfb/if_ogl.c: MFC r43554 |
| 14:37.27 | dli | dloman, then, forget it, it's a 4-year-old core2duo level. |
| 14:37.53 | dloman | heh, i just replaced my old core2 |
| 14:38.09 | dli | dloman, wifi is orders faster than VoIP requires |
| 14:38.37 | dloman | understood |
| 14:39.19 | dloman | and wifi is normally 'good enough' but the wife and I work with large photoshop files and getting them on/off the fileserver via wifi is.... annoying at best. |
| 14:40.28 | dli | dloman, does wifi-N help? |
| 14:40.37 | dloman | some. |
| 14:40.51 | dloman | we're in the process of wiring the house for gig-e |
| 14:41.01 | dloman | so we can plug in when we need speed. |
| 14:41.34 | dli | dloman, that's wonderful :( when I get the budget to redo my home, I will see how to make giga-e possible at home |
| 14:42.29 | dloman | we're doing it slowly. |
| 14:42.31 | dli | dloman, do you work with NURB objects? |
| 14:42.36 | dloman | negative. |
| 14:42.49 | dloman | I know next to nothing about nurbs. |
| 14:42.57 | dloman | its on the long 'tolearn' list. |
| 14:44.32 | dli | dloman, brlcad suggests me to start with NURB intersection problem. I'm too slow here to start :( |
| 14:47.56 | CIA-77 | BRL-CAD: 03erikgreenwald * r43556 10/brlcad/trunk/src/irprep/ir-X.c: size_t casting fixes |
| 14:51.46 | CIA-77 | BRL-CAD: 03erikgreenwald * r43557 10/brlcad/trunk/src/mged/clone.c: size_t casting |
| 14:52.13 | CIA-77 | BRL-CAD: 03erikgreenwald * r43558 10/brlcad/trunk/src/mged/dm-ogl.c: fill out entire struct in table |
| 15:12.03 | brlcad | starseeker: cool, what platform gave you the defined but unused failures? |
| 15:14.00 | brlcad | dloman: you'd saturate the bus before you could drive that many, but it would be fun to see |
| 15:15.05 | brlcad | 4 smaller displays are possible on one dual-dual, or two big'uns |
| 15:15.28 | brlcad | dli: making any progress? |
| 15:15.48 | starseeker | brlcad: was on gentoo |
| 15:15.56 | starseeker | I'm not sure if it was pulling in system itcl or not |
| 15:16.09 | starseeker | those friggin auto_path settings make it problematic |
| 15:16.28 | brlcad | dli: atually the suggestion was based purely on your interest and background -- otherwise, I would have suggested something much smaller to start with :) |
| 15:16.38 | brlcad | starseeker: did it work once installed? |
| 15:16.39 | starseeker | but the gui issue at least was clear - paths were set to build dir, files were in src dir - boom |
| 15:17.02 | starseeker | didn't try an install - was working on regress - but I would expect that it did |
| 15:17.25 | brlcad | k |
| 15:17.29 | brlcad | I'll run a test here too |
| 15:17.47 | brlcad | if install works, then I can at least still tag release |
| 15:18.17 | starseeker | I'm trying to wait for the switch to cmake to really mess with the auto_path settings - they should really simplify down now that we're duplicating the install layout everywhere |
| 15:18.42 | brlcad | true, but it also needs to be relocatable |
| 15:18.54 | starseeker | sure |
| 15:19.05 | brlcad | so an "uninstalled" build tree should also work -- that's the main concern |
| 15:19.06 | starseeker | the CMake build, in my testing, is already relocatable |
| 15:19.42 | brlcad | hm |
| 15:20.05 | starseeker | I just meant we'd have one set of dir paths that list all the tclscripts subdirs, and then use either build root or install root to id them for auto_path |
| 15:21.13 | starseeker | plus, I've got to figure out how to strip those /usr/lib based paths out of the front of the auto_path list |
| 15:22.18 | brlcad | it's not build or install root, it'd be the runtime root for it to be relocatable |
| 15:22.25 | starseeker | er, right |
| 15:22.40 | starseeker | the bu_brlcad_data/bu_brlcad_root logic |
| 15:22.43 | brlcad | k |
| 15:23.28 | brlcad | shouldn't need to strip paths off auto_path -- someone is adding it, that's the point of attack |
| 15:23.58 | starseeker | right - I'm just concerned it might be tcl itself initializing auto_path to those values |
| 15:24.04 | starseeker | at this point I don't know |
| 15:24.11 | brlcad | sounds to me like a system pkgIndex.tcl getting loaded |
| 15:24.52 | starseeker | uh... it can't do that until it has paths to load pkgIndex.tcl files from - that's why it needs that C init process |
| 15:25.23 | brlcad | sure, but there are the built-ini *_Init() routines in the libraries that are loaded |
| 15:25.37 | starseeker | nods |
| 15:25.39 | brlcad | if it pulls a system .so, then it conceivably can modify the auto_path |
| 15:26.36 | starseeker | it would help if there was a debug setting to let us see where package require was finding its .so files |
| 15:26.46 | starseeker | (for a given package require operation) |
| 15:29.02 | brlcad | there are ways to introspect tcl from the interpreter |
| 15:29.15 | brlcad | "info loaded", "into library" |
| 15:29.31 | brlcad | man n info |
| 15:30.15 | starseeker | ah, excellent |
| 15:30.22 | brlcad | "package names" |
| 15:32.05 | starseeker | I have a hunch my gentoo box has a system itcl/itk that happened to work, and it found that (since it didn't have the auto_path information needed to bridge the divide between build and src dirs, based on the gui behavior) |
| 15:32.39 | starseeker | would almost be better if we could somehow reject libraries not in the "BRL-CAD expected" location for Tcl packages |
| 15:32.58 | starseeker | would avoid the accidentally, silently working issue |
| 15:54.58 | CIA-77 | BRL-CAD: 03erikgreenwald * r43559 10/brlcad/trunk/src/mged/ (edsol.c mged.h): de-const due to possible deletion |
| 16:00.21 | CIA-77 | BRL-CAD: 03erikgreenwald * r43560 10/brlcad/trunk/src/mged/fbserv.c: Move the forward declaration into the #if section that uses it, to avoid the "declared 'static' but never defined" error. |
| 16:08.04 | CIA-77 | BRL-CAD: 03erikgreenwald * r43561 10/brlcad/trunk/src/proc-db/vegetation.c: size_t cast fix |
| 16:17.37 | dloman | ``Erik: I think i found it. in GSThread::start(), the pthread was being created in a detached state. In ~GSThread, pthread_join() was called. According to the man page, calling _join on a detached thread is a no no. |
| 16:19.09 | CIA-77 | BRL-CAD: 03davidloman * r43562 10/geomcore/trunk/src/utility/GSThread.cxx: Commented out the pthread_join() call in GSThread::~GSThread. Man page says that a pthread created in a detached state cannot be used as a sync point via pthread_join() |
| 16:19.24 | dloman | see if that messes anything up on bsd and/or mac |
| 16:29.39 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 16:39.39 | CIA-77 | BRL-CAD: 03davidloman * r43563 10/geomcore/trunk/src/utility/GSUuid.cxx: Fix a bug where the buffer used to export the uuid was being free'd prior to that buffer being used to generate a std::string. |
| 17:18.06 | dli | brlcad, yes, I'm still interested on the NURB intersection part. I will have more time now to work on it. Had to go to hospital often in past weeks |
| 17:19.38 | dli | brlcad, at this stage, I'm still reading through the ON wiki: http://wiki.mcneel.com/developer/opennurbs/home |
| 17:20.11 | dli | brlcad, still, I will report my success as well as my failure to you when it's time |
| 17:21.18 | brlcad | naturally, I'd hope we can choose the best development path so we achieve success instead of failur ;) |
| 17:22.02 | brlcad | that might mean not biting on one of the hardest problems first, there are lots of places where some progress could be made that would help get you familiarized with the source code |
| 17:23.16 | dli | brlcad, sure, through evolution is good, but the code base overall is huge, it might easier for me to start by writing something new |
| 17:24.04 | brlcad | new vs old isn't as important as starting with something very small |
| 17:24.25 | brlcad | nurbs intersection is not small ... |
| 17:24.38 | dli | brlcad, if you can point to some place, I would be glad to see whether I can fix some small things in parallel with my intersection problem |
| 17:24.48 | brlcad | so that was probably over-ambitious without having touched any other code |
| 17:26.33 | CIA-77 | BRL-CAD: 03brlcad * r43564 10/brlcad/trunk/doc/README.Linux: include instructions to force 32-bit on platforms that default to 64-bit as well. |
| 17:26.47 | brlcad | hm, something came up just this week that is a nice small task |
| 17:27.23 | brlcad | and it's in a similar section of library code as the brep/nurbs primitive |
| 17:27.28 | brlcad | the revolve primitive |
| 17:28.57 | dli | brlcad, sounds good. is there a bug tracking thread for this task? |
| 17:29.06 | brlcad | basically, it's a very new primitive so it doesn't yet have support for matrix transformations (scaling, translation, rotation) |
| 17:29.46 | brlcad | support for matrix transforms is in one function, which revolve doesn't presently implement |
| 17:30.12 | brlcad | https://sourceforge.net/projects/brlcad/forums/forum/362510/topic/4372998 |
| 17:30.49 | brlcad | that's merely a day or two project, unlike nurbs which is several weeks :) |
| 17:31.34 | dli | brlcad, so, rt_generic_xform(), that's specific enough for me. I will report what I can do later |
| 17:31.53 | brlcad | rt_generic_xform() isn't sufficient, which was my last comment |
| 17:32.13 | brlcad | have to implement rt_revolve_xform() similar to rt_extrude_xform() |
| 17:32.46 | brlcad | src/librt/primitives/revolve/revolve.c and src/librt/primitives/extrude/extrude.c respectively, with the function listed in src/librt/primitives/table.c |
| 17:33.00 | dli | brlcad, I get some rough idea about what's needed here, need to read the code first |
| 17:33.40 | CIA-77 | BRL-CAD: 03brlcad * r43565 10/brlcad/trunk/src/other/tkhtml/Makefile.am: |
| 17:33.40 | CIA-77 | BRL-CAD: the clean rule was overriding autoconf's default clean rule, which calls |
| 17:33.40 | CIA-77 | BRL-CAD: clean-am. this should fix the build where you follow a 32-bit build with a |
| 17:33.40 | CIA-77 | BRL-CAD: 64-bit built and vice versa. stale .o build files were getting left in the |
| 17:33.40 | CIA-77 | BRL-CAD: .libs dir causing the build to halt. |
| 17:33.53 | brlcad | working on that will introduce you to how primitives are implemented, some basic structures, the callback interface, and some light math |
| 17:34.03 | brlcad | all useful and relevant for working on nurbs |
| 17:34.21 | dli | brlcad, let's see. :) |
| 17:38.12 | brlcad | starseeker: do you have a clean build? |
| 17:38.24 | brlcad | ``Erik: when was the last time you tried a windows build? |
| 17:38.33 | brlcad | needs a windows build test |
| 17:38.45 | brlcad | mac/linux are clean here |
| 17:40.24 | brlcad | and ``Erik, what can you tell me about the libtie integration? working on the writeup |
| 17:48.54 | ``Erik | tried one yesterday, broke a lot using msvc8, will try another |
| 17:49.41 | ``Erik | libtie's functions are in librt, bots that are oriented or have normals should pass to tie if you set the rt_min_tie (emulated rt_min_piece) |
| 17:50.13 | *** join/#brlcad Stattrav (~Stattrav@117.192.128.118) |
| 17:50.13 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
| 17:50.15 | ``Erik | the rt_min_tie is set to 4 billion and some change right now pending more testing |
| 17:50.15 | brlcad | how does one set rt_min_tie? |
| 17:50.23 | ``Erik | uh, -c I think? |
| 17:50.36 | brlcad | ah, so it'll auto-kick over to tie for 4M+ models now |
| 17:51.21 | ``Erik | 4m+ bot primitives |
| 17:51.28 | brlcad | right, okay |
| 17:51.32 | ``Erik | 0xffffffff |
| 17:51.39 | brlcad | ah, heh |
| 17:52.09 | brlcad | interesting, so no way to turn it off then or is 0 special? :) |
| 17:52.19 | ``Erik | from when that was an int, not a size_t |
| 17:54.01 | brlcad | and rough performance impact on a 4M model is what? 10%-500%? average 50%? |
| 17:55.06 | ``Erik | didn't have a 4m primitive, at ~2000 it was 200% (100% gain), 200 was like 150% (+50%), 12 was 70% (-30%) |
| 17:56.01 | brlcad | so on a real model, should see the time about cut in half |
| 17:56.43 | ``Erik | that was tesselating a sphere, the 'real' models I tried were tesselations of lots of arbs, not NURBS tesselations from an importer |
| 17:57.03 | ``Erik | the benchmark numbers were from tesselated spheres |
| 17:57.37 | ``Erik | jabs cia |
| 17:58.17 | ``Erik | 0 disables tie as of 6 minutes ago |
| 17:59.01 | brlcad | hehe, awesome |
| 17:59.39 | brlcad | so I'll run a quick test on some bot model I have, see what things look like |
| 18:00.07 | ``Erik | there're a couple rough edges that need cleaned up before I'm comfortable impacting customers with it |
| 18:00.13 | brlcad | any other caveats or useful info other than maybe not so hot for tiny models? |
| 18:00.36 | ``Erik | on occasion, it misses :D |
| 18:00.47 | brlcad | rt_min_tie is coming up empty, what's the real name? :) |
| 18:01.13 | ``Erik | rt_bot_mintie |
| 18:01.23 | brlcad | thx |
| 18:01.24 | ``Erik | (to mimic rt_bot_minpieces) |
| 18:05.36 | CIA-77 | BRL-CAD: 03brlcad * r43566 10/brlcad/trunk/TODO: |
| 18:05.36 | CIA-77 | BRL-CAD: attr command now sorts alphabetically, but then other users reportedly |
| 18:05.36 | CIA-77 | BRL-CAD: also/still want sorting based on creation order (because it makes it easy to |
| 18:05.36 | CIA-77 | BRL-CAD: diff and/or find new additions. need a sorting option. kicking off an EDITOR |
| 18:05.36 | CIA-77 | BRL-CAD: should now be better too now that the invocation has been re-reviewed recently. |
| 18:10.12 | ``Erik | script.c:1414: warning: comparison between signed and unsigned |
| 18:10.19 | ``Erik | src/remrt/remrt.c:752: warning: comparison between signed and unsigned |
| 18:11.10 | CIA-77 | BRL-CAD: 03brlcad * r43567 10/brlcad/trunk/TODO: WE ARE FREE OF COMPILATION WARNINGS! woot. |
| 18:14.51 | CIA-77 | BRL-CAD: 03erikgreenwald * r43568 10/brlcad/trunk/src/librt/primitives/bot/bot.c: rt_bot_mintie=0 now means do not use tie |
| 18:16.43 | brlcad | my 752 is an FD_MOVE line.. doesn't seem right |
| 18:17.11 | ``Erik | I know, I've been grepping around, fbsd might have a bad header |
| 18:17.41 | brlcad | maybe line index from 0 and it's complaining about the tcp_listen_fd int fd |
| 18:19.29 | ``Erik | oohhhhh |
| 18:20.42 | ``Erik | got it |
| 18:21.15 | ``Erik | (#ifndef FD_MOVE ... we had our own for os's like fbsd which don't provide) |
| 18:21.54 | brlcad | interesting |
| 18:24.38 | brlcad | looks like rt_bot_mintie is only exposed via mged tcl var, not -c |
| 18:25.42 | ``Erik | hm, I used rt_bot_minpieces as a template, thought the librt tcl.c was called on rt's startup |
| 18:26.04 | brlcad | i added it |
| 18:26.34 | brlcad | it is a tcl var inside the tcl interp that libtcl has running, but that's not exposed to rt |
| 18:26.53 | brlcad | they are manually wired to the global |
| 18:27.07 | ``Erik | ah |
| 18:27.56 | ``Erik | win32 just built, had to remove regex.h from raytrace.h (or update the include paths for most projects) and add ws2_32.lib to a handful of converters |
| 18:29.55 | brlcad | yeah, regex.h inside raytrace.h doesn't sound like a good idea |
| 18:30.14 | brlcad | should be an implementation detail, not public api requirement |
| 18:31.30 | brlcad | poor CIA-77 |
| 18:31.39 | starseeker | ``Erik: guess you're right, I'll have to do the void thing |
| 18:32.56 | CIA-77 | BRL-CAD: 03erikgreenwald * r43569 10/brlcad/trunk/include/raytrace.h: conditionalize regex.h |
| 18:35.04 | starseeker | brlcad: I'm still getting failures on Mac with script.c from src/tab |
| 18:35.16 | starseeker | script.c:33:5: error: "__STDC_VERSION__" is not defined |
| 18:35.26 | starseeker | script.c: In function âyy_get_next_bufferâ: |
| 18:35.27 | starseeker | script.c:1389: warning: comparison between signed and unsigned |
| 18:40.38 | ``Erik | the windows comment in rt/do.c is due to "initializer not static". might need to assign those in cm_set() as a runtime dealie |
| 18:41.05 | brlcad | aha, that makes more sense |
| 18:41.16 | brlcad | those variables are in librt, so it can't bind them |
| 18:41.22 | CIA-77 | BRL-CAD: 03erikgreenwald * r43570 10/brlcad/trunk/include/raytrace.h: remove regex.h for now, windows would need most vcproj files updated. |
| 18:41.24 | brlcad | dll import suckage |
| 18:46.44 | CIA-77 | BRL-CAD: 03erikgreenwald * r43571 10/brlcad/trunk/src/remrt/remrt.c: fdset uses unsigned, so fix if we need FD_MOVE defined |
| 18:52.33 | CIA-77 | BRL-CAD: 03brlcad * r43572 10/brlcad/trunk/src/rt/do.c: add support for -c rt_bot_mintie |
| 18:53.22 | CIA-77 | BRL-CAD: 03erikgreenwald * r43573 10/brlcad/trunk/misc/win32-msvc8/ (5 files in 5 dirs): link ws2_32.lib for ntohl/htonl |
| 19:00.18 | CIA-77 | BRL-CAD: 03brlcad * r43574 10/brlcad/trunk/src/nirt/ (command.c nirt.c nirt.h): add support for setting rt_bot_mintie from within nirt, similar to rt_bot_minpieces. add new -T option in addition to new bot_mintie nirt command. |
| 19:08.24 | CIA-77 | BRL-CAD: 03brlcad * r43575 10/brlcad/trunk/NEWS: added -T and bot_mintie options to nirt that correspond with controlling when erik's new support for optimized BoT raytacing kicks on |
| 19:12.50 | CIA-77 | BRL-CAD: 03brlcad * r43576 10/brlcad/trunk/src/rt/do.c: expand the comment now that the cause is known. need to set during runtime, not compiletime. |
| 19:13.28 | CIA-77 | BRL-CAD: 03brlcad * r43577 10/brlcad/trunk/misc/win32-msvc/CMakeLists.txt: back to not needing libregex search dir |
| 19:25.46 | CIA-77 | BRL-CAD: 03brlcad * r43578 10/brlcad/trunk/NEWS: (log message trimmed) |
| 19:25.46 | CIA-77 | BRL-CAD: reword tie integration from user-visible perspective. erik integrated the |
| 19:25.46 | CIA-77 | BRL-CAD: former 'libtie' high-performance triangle intersection engine (tie) into librt |
| 19:25.46 | CIA-77 | BRL-CAD: as a way to get optimized BoT raytracing. initial results showing a halfing |
| 19:25.46 | CIA-77 | BRL-CAD: reduction of raytrace time for models at 2k+ triangles. erik added an |
| 19:25.47 | CIA-77 | BRL-CAD: 'rt_bot_mintie' option (exposed via mged and rt -c) for toggling when the |
| 19:25.48 | CIA-77 | BRL-CAD: optimization kicks in. currently set really high at 4M until further testing |
| 19:41.00 | CIA-77 | BRL-CAD: 03starseeker * r43579 10/brlcad/branches/cmake/src/ (libged/search.c librt/search.c): |
| 19:41.01 | CIA-77 | BRL-CAD: Not in final form yet (for one thing, the regex in db_plan_t will have to become |
| 19:41.01 | CIA-77 | BRL-CAD: void and be cast as needed) but this can run searches now. Can't be committed |
| 19:41.01 | CIA-77 | BRL-CAD: to trunk until after release in this form, but committing now in cmake to have |
| 19:41.01 | CIA-77 | BRL-CAD: it checked in somewhere |
| 19:45.23 | CIA-77 | BRL-CAD: 03starseeker * r43580 10/brlcad/branches/cmake/src/libged/search.c: Whoops, need memory here too. |
| 19:52.04 | dloman | brlcad: do we have any Display port to DVI or VGA adapters around the office? |
| 19:52.10 | dloman | (if yo uknow off hand) |
| 19:53.17 | CIA-77 | BRL-CAD: 03starseeker * r43581 10/brlcad/branches/cmake/ (include/raytrace.h src/librt/search.h): Stick regex in the private search header for now... |
| 20:23.41 | starseeker | ``Erik: bob says tcl is wonderful |
| 20:24.22 | ``Erik | especially on windows? |
| 20:25.54 | starseeker | heh |
| 20:51.56 | CIA-77 | BRL-CAD: 03starseeker * r43582 10/brlcad/branches/cmake/ (4 files in 3 dirs): This approach keeps the plan data structure out of raytrace.h, and thus isolates regex. |
| 21:43.55 | CIA-77 | BRL-CAD: 03starseeker * r43583 10/brlcad/branches/cmake/ (23 files in 16 dirs): MFC r43582 |
| 21:48.02 | *** join/#brlcad ``Erik_ (Here@c-69-140-109-104.hsd1.md.comcast.net) |
| 21:58.38 | CIA-77 | BRL-CAD: 03starseeker * r43584 10/brlcad/trunk/ (7 files in 3 dirs): Put the new search routines into trunk |
| 21:59.05 | brlcad | hope you cross-compile tested that :) |
| 22:00.39 | starseeker | brlcad: just mac so far - working on it |
| 22:00.55 | starseeker | had to re-implement the search . ... stuff |
| 22:01.03 | starseeker | I think it behaves the way you wanted it to now |
| 22:01.16 | *** join/#brlcad Klebel (~mk@169.229.55.243) |
| 22:02.11 | brlcad | I actually just hope it works with it being injected right before the release, and isn't like that nirt "fix" made right before release |
| 22:02.32 | starseeker | brlcad: sorry, I know it's a rotten time... |
| 22:02.56 | starseeker | I can back it out in trunk and just work in cmake branch |
| 22:03.49 | CIA-77 | BRL-CAD: 03starseeker * r43585 10/brlcad/branches/cmake/src/ (4 files in 2 dirs): MFC r43584 |
| 22:03.53 | brlcad | committing is fine, you should just be extra care and be testing more than usual |
| 22:04.24 | starseeker | nods - it's hot off the press, I just now got it running, so I'm starting the testing now |
| 22:05.03 | starseeker | if you've got a working compile, you might check if the new behavior of (say) search . -maxdepth=0 does what you expect |
| 22:05.34 | starseeker | I think we can squash that TODO item, but since you spotted the issue confirmation would be good :-) |
| 22:05.58 | brlcad | I can check it (and you should too given the timing) |
| 22:06.04 | starseeker | oh, I am |
| 22:06.41 | starseeker | I just ment I wanted to make sure I had the behavior you wanted, given how hard you had to work to explain it to me :-P |
| 22:07.44 | brlcad | at release time, it becomes more like how trunk development used to be -- trunk HEAD should be treated like stable: changes tested on multiple platforms before commit, runtime tested on everything |
| 22:07.49 | brlcad | I know |
| 22:08.12 | brlcad | I mean "you too" should be making extra sure that all of search still works, maybe run through the documented examples |
| 22:08.20 | brlcad | not just the new feature |
| 22:08.21 | starseeker | ah |
| 22:08.26 | starseeker | gotcha |
| 22:09.50 | starseeker | how |
| 22:09.59 | starseeker | bot.c isn't happy |
| 22:11.13 | brlcad | ah right, warnings |
| 22:11.22 | brlcad | compile had -w in effect |
| 22:11.53 | brlcad | perfect example :) |
| 22:13.34 | starseeker | got those, moving on... |
| 22:15.53 | brlcad | you mean you already got those? |
| 22:17.35 | starseeker | think so - just remove the unused and cast to size_t for comparisons |
| 22:17.53 | starseeker | (don't want to mess with bot->faces... types right now) |
| 22:18.26 | brlcad | have them fixed here too |
| 22:19.21 | starseeker | ah - feel free to stomp mine |
| 22:19.35 | brlcad | yours matched mine |
| 22:19.40 | starseeker | cool |
| 22:19.48 | brlcad | but you apparently weren't getting unused var warnings like I have here |
| 22:20.00 | starseeker | really? you got more? |
| 22:20.05 | brlcad | yep |
| 22:20.09 | starseeker | weird |
| 22:20.25 | brlcad | to be expected |
| 22:21.32 | brlcad | one of the lessons from all the cleanup is that even minor version number differences in gcc result in different warnings, along with changes to compilation options, 32-bit vs 64-bit, optimized vs non-optimized |
| 22:22.24 | brlcad | and they're not a combination that is exactly a superset, so there ends up being something like 3! possible configurations |
| 22:22.35 | brlcad | 3! or 4! |
| 22:25.04 | starseeker | nods |
| 22:30.06 | _psilva | :( |
| 22:30.21 | starseeker | _psilva: ? |
| 22:35.17 | starseeker | brlcad: my mac compile is still failing in src/tab on script.c |
| 22:37.14 | starseeker | search completes all the examples from the man page on potential |
| 22:38.23 | starseeker | brlcad: what version of flex and bison (or lex/yacc) are you working with? |
| 22:55.22 | starseeker | oh, right |
| 22:55.34 | starseeker | can't really test well on the mac because of that messed up install |
| 22:59.44 | starseeker | with autotools anyway... |
| 22:59.57 | starseeker | search looks ok with the cmake build on the mac |
| 23:16.15 | CIA-77 | BRL-CAD: 03starseeker * r43586 10/brlcad/trunk/src/libged/search.c: Re-add support for the '.' option (e.g. search . -name s*) but this time do it at the ged level with post-processing of the full search. Also doesn't print the leading '/' character for the '.' searches. |
| 23:17.27 | CIA-77 | BRL-CAD: 03brlcad * r43587 10/brlcad/trunk/src/conv/patch/patch-g.c: curious that thick_no only shows up as a potential longjmp clobber var when compiling in 32-bit mode |
| 23:17.47 | CIA-77 | BRL-CAD: 03brlcad * r43588 10/brlcad/trunk/src/librt/primitives/bot/bot.c: damnits, found a bug during release testing. disable the optimization so release can proceed. |
| 23:17.51 | CIA-77 | BRL-CAD: 03brlcad * r43589 10/brlcad/trunk/TODO: resolve the crash post-release |
| 23:22.18 | CIA-77 | BRL-CAD: 03starseeker * r43590 10/brlcad/trunk/src/librt/primitives/bot/bot.c: Clear some warnings on bot.c |
| 23:24.29 | CIA-77 | BRL-CAD: 03brlcad * r43591 10/brlcad/trunk/src/librt/primitives/bot/bot.c: quell unused var warnings for the non-optimized case. split vars across impls. |
| 23:26.07 | _psilva | gdc crunch sucks |
| 23:26.45 | _psilva | need more comp days from this |
| 23:32.00 | *** join/#brlcad Klebel (~mk@w72.RIC.Berkeley.EDU) |
| 23:53.03 | brlcad | sushi:~ morrison$ flex --version |
| 23:53.03 | brlcad | flex 2.5.35 |
| 23:53.03 | brlcad | sushi:~ morrison$ bison --version |
| 23:53.03 | brlcad | bison (GNU Bison) 2.3 |