| 00:08.46 | CIA-28 | BRL-CAD: 03starseeker * r48006 10/brlcad/trunk/CMakeLists.txt: Need to do the 32/64 bit check earlier in the game, before we try to find anything at all. |
| 00:09.41 | CIA-28 | BRL-CAD: 03n_reed * r48007 10/brlcad/trunk/src/other/perplex/ (scanner.re scanner_template.c): Comparing disparate pointers is undefined. Don't do it. |
| 00:37.42 | CIA-28 | BRL-CAD: 03n_reed * r48008 10/brlcad/trunk/src/other/perplex/ (scanner.re scanner_template.c): need to update input markers if input buffer is reallocated |
| 01:55.12 | CIA-28 | BRL-CAD: 03starseeker * r48009 10/brlcad/trunk/ (10 files in 2 dirs): Slew of documentation updates. Not claiming final form yet, but closer. |
| 02:07.49 | CIA-28 | BRL-CAD: 03starseeker * r48010 10/brlcad/trunk/doc/README.Windows: Start reworking the Windows docs for CMake. |
| 02:13.23 | CIA-28 | BRL-CAD: 03starseeker * r48011 10/brlcad/trunk/ (8 files): Move the CMake versions of doc files into place. |
| 02:19.18 | dli | svn version building error: http://pastebin.com/HUC6ibBc |
| 02:23.58 | CIA-28 | BRL-CAD: 03starseeker * r48012 10/brlcad/trunk/ (INSTALL src/other/CMakeLists.txt): Copy/paste strikes again |
| 02:33.51 | CIA-28 | BRL-CAD: 03starseeker * r48013 10/brlcad/trunk/configure.cmake.sh: Add a bunch more of the options to the configure script. Looking more like it will be possible to simply autogenerate the verbose part of this. Also points out there are a few toplevel options that still need docs. |
| 02:34.39 | starseeker | dli: can you paste it using the mozilla pastebin? |
| 02:34.52 | dli | starseeker, ok |
| 02:35.40 | dli | starseeker, http://pastebin.mozilla.org/1407174 |
| 02:39.59 | starseeker | um... not sure what's going on - I just got a clean build from latest trunk here... what were your configure options? |
| 02:41.58 | dli | starseeker, one moment |
| 02:45.05 | dli | starseeker, http://pastebin.mozilla.org/1407179 |
| 02:45.58 | dli | starseeker, reported from cmake: http://pastebin.mozilla.org/1407180 |
| 02:46.34 | starseeker | holy bleep |
| 02:46.43 | starseeker | are you building it as part of a packaging system? |
| 02:46.56 | dli | starseeker, yes, in gentoo |
| 02:46.59 | starseeker | oh, I see gentoo |
| 02:47.21 | starseeker | well, for one thing, a lot of the options are incorrect for trunk |
| 02:48.23 | starseeker | and they shouldn't be trying to turn off scl or utahrle, unless they've got ebuilds of our own src/other srcs |
| 02:48.39 | starseeker | dli: where did you get that ebuild from? |
| 02:48.43 | starseeker | is that an overlay? |
| 02:48.48 | dli | starseeker, yes |
| 02:48.53 | starseeker | which one? |
| 02:49.06 | dli | starseeker, science-overlay |
| 02:49.50 | starseeker | hmm. I'd have to take a look, but it looks like someone either isn't maintaining that ebuild or isn't paying close enough attention |
| 02:50.09 | dli | starseeker, very old ebuilds |
| 02:54.30 | starseeker | yeah, this needs a total overhaul |
| 02:54.36 | starseeker | who's maintaining it? |
| 02:55.37 | starseeker | he DISABLED opengl?? |
| 02:56.03 | starseeker | and did it wrong to boot - it's -DBRLCAD_ENABLE_OPENGL=OFF, not -BRLCAD_ENABLE_OPENGL=OFF |
| 02:56.06 | dli | starseeker, I disabled opengl, because builds fails with opengl |
| 02:56.35 | starseeker | O.o - why? |
| 02:57.17 | starseeker | dli: can you do me a favor? Just grab a trunk checkout and build it "stand-alone" without the ebuild? |
| 02:57.35 | starseeker | cmake .. -DBRLCAD_BUNDLED_LIBS=ON and see what happens? |
| 02:57.50 | dli | starseeker, sure |
| 03:00.17 | starseeker | all the BRLCAD_BUILD_LOCAL variables are obsolete and not doing anything |
| 03:00.49 | dli | starseeker, reported from cmake: http://pastebin.mozilla.org/1407198 |
| 03:00.58 | starseeker | to achieve what it looks like he's after, all he had to do was -DBRLCAD_BUNDLED_LIBS=System , but that doesn't stand a good chance of working... |
| 03:01.18 | starseeker | dli: try building once |
| 03:01.40 | dli | starseeker, it's building |
| 03:02.25 | dli | starseeker, /var/tmp/portage/media-gfx/brlcad-9999/work/brlcad-9999/include/sysv.h:61:26: error: conflicting types for ‘strchr’ |
| 03:02.58 | dli | starseeker, that's from cmake -DBRLCAD_BUNDLED_LIBS=ON . |
| 03:04.58 | starseeker | what's the full error? |
| 03:05.10 | starseeker | previous definition? |
| 03:06.17 | dli | starseeker, that's all I got: http://pastebin.mozilla.org/1407203 |
| 03:06.32 | starseeker | also, you're still using the gentoo brlcad-9999 sources? |
| 03:07.41 | dli | starseeker, should not be, I run cmake in the source folder, gentoo builds it in a different folder |
| 03:07.56 | dli | starseeker, I can try to copy the source and try again |
| 03:09.20 | starseeker | dli: I'd suggest starting totally clean, if you can |
| 03:09.34 | dli | starseeker, no problem |
| 03:10.09 | starseeker | in your home directory: svn co https://brlcad.svn.sf.net/svnroot/brlcad/brlcad/trunk brlcad && mkdir brlcad-build && cd brlcad-build && cmake ../brlcad -DBRLCAD_BUNDLED_LIBS=ON && make |
| 03:10.38 | starseeker | clearly turning off opengl in CMake is causing some... unexpected issues, I'll have to look into it |
| 03:11.59 | dli | starseeker, trying again |
| 03:17.49 | dli | starseeker, better, that error doesn't show up |
| 03:19.07 | CIA-28 | BRL-CAD: 03starseeker * r48014 10/brlcad/trunk/misc/CMake/ThirdParty.cmake: Correct a few gross errors in ThirdParty.cmake, but still not clear how disabling opengl is messing with everything. |
| 03:19.41 | starseeker | I gotta go - dli, post how that turns out. Clearly I've got a bug to fix, but that ebuild is hopelessly out of date |
| 03:20.03 | dli | starseeker, ok |
| 04:43.57 | CIA-28 | BRL-CAD: 03starseeker * r48015 10/brlcad/trunk/misc/CMake/ (ThirdParty.cmake ThirdParty_TCL.cmake): Er, right, macros. Reset a conditional variable before using it again... |
| 05:01.34 | starseeker | dli: did it build? |
| 05:03.12 | dli | starseeker, it builds in a fresh folder |
| 05:03.25 | dli | starseeker, but still couldn't get it build in ebuild |
| 05:05.30 | starseeker | that ebuild's settings are way too aggressive |
| 06:11.59 | CIA-28 | BRL-CAD: 03starseeker * r48016 10/brlcad/trunk/ (3 files in 3 dirs): |
| 06:12.01 | CIA-28 | BRL-CAD: Do more work to make BRLCAD_BUNDLED_LIBS=System do what it claims to do. Also |
| 06:12.01 | CIA-28 | BRL-CAD: make Tcl packages respect it, unless BRLCAD_TCL is overridden locally - then |
| 06:12.01 | CIA-28 | BRL-CAD: respect the local Tcl setting rather than BUNDLED_LIBS. Still needs some |
| 06:12.01 | CIA-28 | BRL-CAD: testing - not a terribly useful setting for us, but the Linux distributions will |
| 06:12.01 | CIA-28 | BRL-CAD: probably want it so we should shake it out and make it behave. |
| 06:12.40 | CIA-28 | BRL-CAD: 03starseeker * r48017 10/brlcad/trunk/INSTALL: Doing Tk as a Tcl extension, so docs are generated for it - add 'em. |
| 06:48.24 | *** join/#brlcad jordisayol (~jordisayo@unaffiliated/jordisayol) |
| 10:33.02 | *** join/#brlcad hackrat (~packrator@c-98-209-146-133.hsd1.mi.comcast.net) |
| 11:54.46 | *** join/#brlcad ``Erik (~erik@BRLCAD.ORG) |
| 14:12.25 | *** join/#brlcad packrat (~packrator@c-98-209-146-133.hsd1.mi.comcast.net) |
| 16:22.42 | CIA-28 | BRL-CAD: 03starseeker * r48018 10/brlcad/trunk/src/other/CMakeLists.txt: Needing the C library from a Tcl/Tk package is compilcating things... however, a bit of a pattern is emerging... |
| 16:27.11 | dli | starseeker, I updated ebuilds in gentoo science overlay. Noticed building failures due to: LDFLAGS "--as-needed", BRLCAD_ENABLE_RTSERVER=ON, and CXXFLAGS="-std=c++0x" |
| 16:32.28 | starseeker | hmm. yeah, rtserver's build can be a bit finicky |
| 16:32.44 | starseeker | that's the java bit - I doubt most people would miss it |
| 16:33.11 | starseeker | hehe - this is awesome: |
| 16:33.23 | starseeker | '"Computer scientists" don't write code, they write books explaining why all existing implementations are wrong.' |
| 16:34.58 | dli | starseeker, still, gentoo wants features be specified by USE flags, not but auto-detection, so, it's not encouraged to use BRLCAD_BUNDLED_LIBS=AUTO |
| 16:42.23 | starseeker | dli: yeah, I know - one of the longer active bugs in the history of the gentoo bug database was the "here's a BRL-CAD ebuild" discussion arguing about things like autodetection and default installation location |
| 16:43.11 | dli | starseeker, default location settled down to /usr/brlcad/, iirc |
| 16:43.16 | starseeker | The commits last night were working to fix the "System" setting for BRLCAD_BUNDLED_LIBS. That's what distributions *should* set in the future if they don't want any of our bundled libs |
| 16:44.04 | starseeker | It will still annoy them because System won't disable opennurbs, scl, and one or two other src/other libraries |
| 16:44.32 | starseeker | the reason for that, however, is those are *locally modified* copies - a system version, at least at the moment, would not be expected to work correctly |
| 16:45.39 | starseeker | we *could* stick those local "forks" into our own source tree and avoid the issue, but the point/hope is to eventually establish those as separate projects - progress is already being made in that regard with scl |
| 16:46.10 | starseeker | src/other libraries and tools tend to have more uses than just BRL-CAD, so there is a reasonable hope they will attract other interest |
| 16:46.47 | starseeker | the odds of that go up if it's easy to build the library "in isolation" |
| 16:47.38 | starseeker | so we maintain our build system in such a way that those libraries are seaprate projects, even though we're the only users/source for those particular versions |
| 16:47.51 | starseeker | s/seaprate/separate/g |
| 16:50.28 | starseeker | scl (the Step Class Libraries) is showing activity on a github project, and once we get things to the point where their build and auto-generated output work for us and support our step-g convertor, we'll be able to use a hypothetical system SCL install |
| 16:51.10 | dli | starseeker, setting to System fails with svn: http://pastebin.mozilla.org/1408136 |
| 16:51.54 | starseeker | dli: do you have itcl and itk installed? |
| 16:52.38 | dli | starseeker, yes: Installed versions: 3.4_beta1, Installed versions: 3.4_pre20090417-r1 |
| 16:53.05 | starseeker | where is libitcl located? |
| 16:53.11 | starseeker | and libitk? |
| 16:53.39 | dli | starseeker, /usr/lib64/itcl3.4/libitcl3.4.so |
| 16:53.53 | dli | /usr/lib64/itk3.4/libitk3.4.so |
| 16:54.23 | starseeker | alright... let me see if I can tell why the find_library calls aren't seeing those |
| 16:56.13 | starseeker | oh, by the way - unless gentoo's lemon package puts lempar.c in /usr/bin/, lemon isn't going to work |
| 16:59.45 | CIA-28 | BRL-CAD: 03starseeker * r48019 10/brlcad/trunk/src/other/CMakeLists.txt: We need the ITCL version set for find_library... since we aren't testing for it when we're going whole-hog system, set it to 3.4 if it's not defined. |
| 17:00.44 | starseeker | need to think about that one... ideally, even if we're forcing system we'd still want the version installed if there is one... |
| 17:01.10 | starseeker | jeez, this business of using Tcl/Tk package C libraries is a nightmare |
| 17:07.58 | starseeker | actually, gentoo doesn't even have its own lemon package - must be part of sqlite |
| 17:09.33 | starseeker | aaand the installed one doesn't like something |
| 17:09.39 | starseeker | even with lempar.c present |
| 17:14.31 | starseeker | dli: you have re2c installed? |
| 17:14.52 | starseeker | dev-util/re2c-0.13.5 is what I've got here... |
| 17:15.03 | dli | starseeker, yes, Installed versions: 0.13.5 |
| 17:15.07 | starseeker | ok |
| 17:15.24 | starseeker | that looks like it should work, but the system lemon is a disaster |
| 17:15.35 | starseeker | it's not even versioned on its own |
| 17:16.21 | starseeker | and I can't stick lempar.c in /usr/bin even if this version HAD worked... talk about your sandbox violations |
| 17:16.47 | starseeker | perplex is written by one of our own developers to wrap re2c, so there won't be a package for that one |
| 17:16.51 | starseeker | one sec... |
| 17:19.25 | CIA-28 | BRL-CAD: 03starseeker * r48020 10/brlcad/trunk/ (misc/CMake/ThirdParty.cmake src/other/CMakeLists.txt): (log message trimmed) |
| 17:19.25 | CIA-28 | BRL-CAD: Extend the NOSYS mechanism to tools as well. Perplex is our own tool, so even a |
| 17:19.25 | CIA-28 | BRL-CAD: SYSTEM install isn't going to have it available (it's our own code, just being |
| 17:19.25 | CIA-28 | BRL-CAD: treated as a separate project) and trying to use a system version of lemon is... |
| 17:19.25 | CIA-28 | BRL-CAD: problematic, to say the very least. lempar.c isn't likely to be present, and |
| 17:19.25 | CIA-28 | BRL-CAD: even when our copy is placed in /usr/bin problems occurred. lemon isn't |
| 17:19.25 | CIA-28 | BRL-CAD: packaged by itself, so it's hard to know what to do about that... suggest |
| 17:22.06 | CIA-28 | BRL-CAD: 03starseeker * r48021 10/brlcad/trunk/src/libged/simulate/simrt.c: shadowed variable warning... |
| 17:29.48 | *** join/#brlcad Elrohir (~kvirc@p579F0159.dip.t-dialin.net) |
| 17:33.32 | dli | starseeker, it works now, with System |
| 17:34.08 | starseeker | There's a lemon ebuild in an overlay: http://gpo.zugaina.org/dev-util/lemon/euscan |
| 17:37.55 | CIA-28 | BRL-CAD: 03n_reed * r48022 10/brlcad/trunk/src/other/perplex/ (scanner.re scanner_template.c): cleanup |
| 17:38.09 | starseeker | that DOES work, if I re-work the FindLEMON logic a bit |
| 17:38.25 | starseeker | http://gentoo-overlays.zugaina.org/gentoo-zh/portage/dev-util/lemon/ |
| 17:38.49 | starseeker | that would be a good one to advocate for in the main gentoo trunk, to help a system BRL-CAD build work |
| 17:39.06 | starseeker | he's patched lemon so the template file can live in /usr/share |
| 17:52.49 | CIA-28 | BRL-CAD: 03starseeker * r48023 10/brlcad/trunk/ (misc/CMake/FindLEMON.cmake src/other/CMakeLists.txt): Redo the handling of lemon a bit, since there does exist at least one working path for a viable system lemon installation. |
| 17:54.18 | starseeker | hah - there's a redhat package too, that apparently does the same thing |
| 17:54.20 | starseeker | sweet |
| 17:54.28 | starseeker | http://rpm.pbone.net/index.php3/stat/4/idpl/15156214/dir/redhat_el_5/com/lemon-3.6.20-1.el5.i386.rpm.html |
| 18:00.11 | starseeker | and it looks like FreeBSD does something sane |
| 18:00.18 | starseeker | looks like Gentoo is behind the curve |
| 18:00.24 | starseeker | embarassing on a dev tool |
| 18:03.18 | CIA-28 | BRL-CAD: 03starseeker * r48024 10/brlcad/trunk/src/other/CMakeLists.txt: Rework explanatory text for lemon |
| 18:04.51 | starseeker | dli: it should still work with Gentoo, but you'll probably need that overlay ebuild install of lemon |
| 18:05.04 | starseeker | or specify -DBRLCAD_LEMON=Bundled |
| 18:09.28 | starseeker | and there's no tkhtml ebuild that I can see, so out of the box archer won't work... |
| 18:18.25 | CIA-28 | BRL-CAD: 03starseeker * r48025 10/brlcad/trunk/misc/CMake/ThirdParty_TCL.cmake: Add a note about another refactor needed to allow local enabling - right now can't set SYSTEM and then turn on just Tkhtml |
| 19:39.56 | CIA-28 | BRL-CAD: 03bob1961 * r48026 10/brlcad/trunk/src/libtclcad/tclcad_obj.c: Allow empty polygons. |
| 19:45.46 | CIA-28 | BRL-CAD: 03bob1961 * r48027 10/brlcad/trunk/src/tclscripts/lib/Ged.tcl: Add callbacks for polygon creation. |
| 21:50.30 | CIA-28 | BRL-CAD: 03n_reed * r48028 10/brlcad/trunk/misc/CMake/FindPERPLEX.cmake: Corrected perplex and re2c options. Basing re2c intermediate name on output name, not output path. |
| 22:46.09 | CIA-28 | BRL-CAD: 03n_reed * r48029 10/brlcad/trunk/src/libgcv/wfobj/tri_face.c: s/BU_GET/bu_pool_get/ so that freeing with standard nmg routines doesn't cause bomb. |