00:28.45 | brlcad | starseeker: I just finished a clean recompile to test |
00:28.48 | brlcad | it's working |
00:29.01 | brlcad | the problem is running tcl applications without having installed BRL-CAD first |
00:29.21 | brlcad | have to make install BEFORE make test or it'll fail on OS X .. haven't figured out how to fix that yet |
00:30.17 | brlcad | it finds the system-installed Tcl at runtime |
00:31.37 | starseeker | ah |
00:32.21 | starseeker | yeah, I'll usually test in CMake if I'm doing a non-installed trial.. |
00:35.03 | brlcad | it should fail in cmake just the same actually unless they modify the LD paths accordingly |
00:36.33 | starseeker | I've got some RPATH settings enabled that seem to do wonders |
00:37.24 | starseeker | you should give it a try sometime :-P |
00:40.08 | brlcad | oh, I believe ya -- another plus if they have better rpath management |
00:41.12 | CIA-53 | BRL-CAD: 03brlcad * r42934 10/brlcad/trunk/m4/patch.m4: (log message trimmed) |
00:41.13 | CIA-53 | BRL-CAD: sigh, no response from the libtool mailing list on this issue so restore the |
00:41.13 | CIA-53 | BRL-CAD: hack that makes libtool binary wrapper scripts work on Mac OS X. we sneak in |
00:41.13 | CIA-53 | BRL-CAD: paths to the Tcl/Tk compile-time lib directories so that they'll get |
00:41.13 | CIA-53 | BRL-CAD: automatically added to DYLD_LIBRARY_PATH. due to suck in the way temp_rpath is |
00:41.13 | CIA-53 | BRL-CAD: handled, the trailing ':' is required otherwise the path will get munged with |
00:41.14 | CIA-53 | BRL-CAD: the next pathlist incorrectly. document why this oddity is even in here, why |
00:42.50 | starseeker | brlcad: about the --enable-only-benchmark option... does that actively squash other options and turn on just what's needed for the benchmark build? |
00:43.04 | starseeker | (also, what's the advantage of that over a normal configure and make benchmark?) |
00:45.09 | brlcad | it only enables and builds exactly enough to run the benchmark |
00:46.06 | brlcad | greatly shortens the build time (by less than a 10th iirc) |
00:46.52 | brlcad | reduces the configure tests and reduces compilation to approximately the minimum |
00:47.37 | starseeker | so the configure time is the difference between that and just doing a normal configure + make benchmark? |
00:48.55 | brlcad | make benchmark still builds everything, then runs cd bench && make bench |
00:49.15 | brlcad | "everything" for an --enable-only-benchmark build is the bare minimum |
00:49.26 | brlcad | mged, asc2g, pixcmp, etc |
00:49.26 | starseeker | ah. |
00:50.00 | starseeker | The CMake build is set up so that a normal configure + make benchmark will build just what is needed for the benchmark and run it - is this an acceptable alternative? |
00:50.13 | brlcad | in theory, with cmake you should be able to describe a new benchmark rule that has the proper minimal dependencies |
00:50.22 | starseeker | did :-) |
00:50.29 | brlcad | yep, that's fine |
00:50.34 | starseeker | sweet |
00:51.12 | brlcad | the only difference is probably that "make install" would only install the benchmark too :) |
00:51.19 | ``Erik | less than 1/10th... before opennurbs? :) |
00:51.54 | brlcad | heh, don't know |
00:52.13 | starseeker | is betting yes - librt needs opennurbs, which is a large percentage of time |
00:52.40 | brlcad | still gets to avoid step dir, half of src/other, hundreds of other utils |
00:52.58 | starseeker | true |
00:53.00 | ``Erik | um, it blindly does src/conv which has the step stuff, right? |
00:53.13 | brlcad | don't recall |
00:53.18 | brlcad | it might |
00:53.33 | starseeker | well, make benchmark is pretty quick in CMake - that does avoid step |
00:53.44 | starseeker | but there's no escaping opennurbs :-P |
00:53.57 | brlcad | easy enough for someone to test the difference if it mattered |
00:54.20 | brlcad | starseeker: you have rough idea of how long a full build and benchmark-only build take? |
00:57.32 | starseeker | with cmake or autotools? |
00:57.55 | starseeker | also, benchmark skips all the static libs which makes a big difference |
00:58.21 | brlcad | either |
00:58.23 | brlcad | both |
00:58.29 | brlcad | <PROTECTED> |
00:58.29 | brlcad | <PROTECTED> |
00:58.52 | brlcad | ouch.. that's a sad v4 flip case |
00:59.18 | brlcad | 1 plithy object that failed to validate whether flipped or not causing trouble for everything else |
00:59.24 | starseeker | brlcad: let me do some tests... |
00:59.31 | starseeker | ouch |
00:59.43 | starseeker | that reminds me - does the flip routine handle ars primitives? |
01:00.02 | brlcad | I talked to D, that's what I'm looking into |
01:01.09 | starseeker | ah, cool |
01:03.02 | *** join/#brlcad yukonbob (~bch@S010600235a187d92.ok.shawcable.net) |
01:14.14 | starseeker | CMake make -j3 benchmark, from beginning to start of actually running the benchmarks, not including configure time - ~2 minutes, maybe a bit less |
01:18.33 | starseeker | hmm - pulled in libpc on autotools... |
01:20.57 | starseeker | little over 4 minutes for the same thing with autotools |
01:21.24 | starseeker | appears it did not pull in step in either case |
01:22.04 | starseeker | both of those times used system libs instead of building local copies |
01:32.46 | brlcad | starseeker: technically, librt needs libpc |
01:33.13 | starseeker | 8.5 minutes for an autotools build, no docbook |
01:33.44 | brlcad | full cmake build takes how long? |
01:33.44 | starseeker | brlcad: ah - I haven't set up that dependency in CMake yet - not a problem, just haven't done so yet |
01:33.51 | starseeker | trying that next |
01:34.00 | starseeker | (usually leave docbook on, messes with numbers) |
01:34.10 | brlcad | sure |
01:34.57 | CIA-53 | BRL-CAD: 03brlcad * r42935 10/brlcad/trunk/regress/fastgen.sh: add diagnostic 'command prompts' to let you know what the important steps being taken are |
01:36.06 | starseeker | again, using system libs for all builds (not enable-all) |
01:36.20 | *** join/#brlcad juanman (~quassel@unaffiliated/juanman) |
01:43.34 | starseeker | about 6.5 minutes for the cmake build |
01:45.15 | starseeker | so about half the time for a benchmark build (once I throw libpc into the CMake build) |
01:46.02 | starseeker | maybe a bit less |
01:46.35 | starseeker | actually, the autotools build reports exact time - 8 min, 33 seconds |
01:46.49 | starseeker | looks at CMake build... |
01:56.02 | starseeker | ah right, only got it going for configure... |
01:59.05 | brlcad | you don't calculate compilation time?! :) |
02:02.26 | brlcad | perhaps ironic, but I actually think cleanly and clearly reporting build time is very important for a variety of reasons |
02:03.55 | brlcad | passive vs active metrics, interface feedback, completion confirmation |
02:11.28 | starseeker | brlcad: it's a little tricky... working on it now |
02:28.22 | *** join/#brlcad yukonbob (~bch@S0106002129e399fc.ok.shawcable.net) |
02:29.43 | CIA-53 | BRL-CAD: 03brlcad * r42936 10/brlcad/trunk/include/raytrace.h: note the three unused debug bits remaining |
02:34.31 | CIA-53 | BRL-CAD: 03brlcad * r42937 10/brlcad/trunk/src/lgt/lgt.h: lgt should not replicate raytrace.h debug values |
02:35.42 | CIA-53 | BRL-CAD: 03brlcad * r42938 10/brlcad/trunk/ (include/raytrace.h src/librt/cut.c src/librt/prep.c): rename DEBUG_PLOTSOLIDS and DEBUG_PLOTBOX to DEBUG_PL_SOLIDS and DEBUG_PL_BOX respectively just because it's 1-char shorter and matches nmg.h |
02:51.56 | CIA-53 | BRL-CAD: 03brlcad * r42939 10/brlcad/trunk/src/librt/ (db_corrupt.c librt.3): (log message trimmed) |
02:51.56 | CIA-53 | BRL-CAD: the user needs SOME mechansim to override automatic behavior in case there is |
02:51.56 | CIA-53 | BRL-CAD: some pressing need, so provide for a LIBRT_V4FLIP boolean environment variable. |
02:51.56 | CIA-53 | BRL-CAD: if set, the value will be interpreted as an override to enable or disable |
02:51.56 | CIA-53 | BRL-CAD: flipping so you can flip a file irrespective of the automatic detection. |
02:51.56 | CIA-53 | BRL-CAD: setting the flag to false will revert to old behavior. setting to true will |
02:51.57 | CIA-53 | BRL-CAD: flip even if the application provided no override and automatic detection would |
02:54.39 | CIA-53 | BRL-CAD: 03starseeker * r42940 10/brlcad/branches/cmake/CMakeLists.txt: Remove the benchmark and rtserver-only options - benchmark is handled by proper dependencies for make benchmark. rtserver will also build minimally due to dependencies. |
03:11.09 | CIA-53 | BRL-CAD: 03starseeker * r42941 10/brlcad/branches/cmake/src/CMakeLists.txt: Without the build options for benchmark and framework, the src/CMakeLists.txt logic simplifies down nicely |
03:13.58 | CIA-53 | BRL-CAD: 03starseeker * r42942 10/brlcad/branches/cmake/ (4 files in 2 dirs): |
03:13.58 | CIA-53 | BRL-CAD: Add a mechanism to print out the compilation time for CMake build. Works only |
03:13.58 | CIA-53 | BRL-CAD: for make (e.g. make all) which appears to be consistent with autotools. Summary |
03:13.58 | CIA-53 | BRL-CAD: information isn't as elaborate as the autotools output, but that can be added if |
03:13.58 | CIA-53 | BRL-CAD: it's needed. |
03:14.07 | starseeker | brlcad: there we go |
03:48.56 | brlcad | example output? |
03:49.25 | brlcad | omg, this is delicious .. |
03:49.42 | brlcad | spam musubi |
03:49.53 | brlcad | "hawaiian sushi" |
04:17.06 | CIA-53 | BRL-CAD: 03brlcad * r42943 10/brlcad/trunk/configure.ac: still having woes with debugging, perhaps the flip was to gstabs+ for 10.4+? |
04:52.44 | *** join/#brlcad Ralith (~ralith@S010600221561996a.vc.shawcable.net) |
05:21.34 | *** join/#brlcad Stattrav (~Stattrav@122.172.33.253) |
05:21.34 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) |
06:51.29 | CIA-53 | BRL-CAD: 03Sean 07http://brlcad.org * r2458 10/wiki/Backstaff: backstaff beginnings |
06:58.41 | CIA-53 | BRL-CAD: 03brlcad * r42944 10/brlcad/trunk/src/librt/db_open.c: past tense on endianness flippage |
07:03.35 | CIA-53 | BRL-CAD: 03brlcad * r42945 10/brlcad/trunk/src/librt/db_corrupt.c: |
07:03.35 | CIA-53 | BRL-CAD: be a lot more eager to flip the endian interpretation of v4 files now that |
07:03.35 | CIA-53 | BRL-CAD: there's an environment variable override. if we merely detect that it'll help a |
07:03.35 | CIA-53 | BRL-CAD: majority of objects, let the flip occur. Moreover, report all objects |
07:03.35 | CIA-53 | BRL-CAD: containing matrices that are invalid regardless of flipping since they may just |
07:03.35 | CIA-53 | BRL-CAD: actually be invalid. summarize how many objects are like that. |
07:13.07 | CIA-53 | BRL-CAD: 03brlcad * r42946 10/brlcad/trunk/TODO: confirmed that ARS has a problem importing because it manages its own b-record serialization of dbfloat_t data. looks like old POLY objects will have a similar problem too. opendb is now fixed. |
07:18.59 | CIA-53 | BRL-CAD: 03brlcad * r42947 10/brlcad/trunk/src/librt/primitives/ars/ars.c: there is no reason for rt_ars_rd_curve() to be public api. rename sans rt_ prefix and mark HIDDEN. |
07:30.54 | CIA-53 | BRL-CAD: 03brlcad * r42948 10/brlcad/trunk/include/db.h: B_solid records don't actually seem to be used anywhere anymore. should be safe to remove from the union. |
07:36.57 | CIA-53 | BRL-CAD: 03brlcad * r42949 10/brlcad/trunk/include/db.h: apparently not true. nothing refers to struct B_solid, but the bspline primitive does refer to and use the B union record. revert. |
07:41.20 | CIA-53 | BRL-CAD: 03brlcad * r42950 10/brlcad/trunk/include/db.h: rename the B_resolution confusion starting point as it doesn't look to be actually used (at least by bspline). update comment and name, however, instead of changing the struct size just in case that matters. |
07:43.41 | CIA-53 | BRL-CAD: 03brlcad * r42951 10/brlcad/trunk/TODO: that means bsplines won't auto-flip either |
08:01.54 | *** join/#brlcad epileg (~epileg@unaffiliated/epileg) |
08:14.51 | *** join/#brlcad d_rossberg (~rossberg@BZ.BZFLAG.BZ) |
08:22.05 | d_rossberg | starseeker: because of the CMake build: i set BRLCAD_PREFIX to something (in the GUI) but "Prefix:" is empty in the configuration summary |
09:07.17 | *** join/#brlcad mafm_ (~mafm@193.153.53.189) |
13:06.05 | *** join/#brlcad juanman (~quassel@unaffiliated/juanman) |
13:21.28 | starseeker | d_rossberg: you shouldn't need to use BRLCAD_PREFIX for anything anymore |
13:21.44 | starseeker | CMAKE_INSTALL_PREFIX should be respected now |
13:27.03 | starseeker | brlcad: http://pastebin.mozilla.org/1018272 |
13:27.09 | starseeker | very simple right now |
13:27.43 | d_rossberg | there is no CMAKE_INSTALL_PREFIX declared, i can't access it via the CMake GUI |
13:28.03 | starseeker | um |
13:28.11 | starseeker | this is on Windows? |
13:29.58 | starseeker | updates his copy on windows to the latest... |
13:30.31 | CIA-53 | BRL-CAD: 03d_rossberg * r42952 10/brlcad/trunk/include/common.h: there are unused parameters which are asserted |
13:30.45 | d_rossberg | the cmake gui should be similar on all systems |
13:31.08 | starseeker | yes, it should |
13:31.28 | starseeker | but there is conditional logic for path setup |
13:31.49 | starseeker | is checking |
13:32.05 | starseeker | I usually run from the command line, so it's possible I missed something |
13:32.45 | d_rossberg | following my documentation CMAKE_INSTALL_PREFIX isn't a standard variable but CMAKE_ROOT is |
13:35.28 | starseeker | uh... http://www.cmake.org/cmake/help/cmake-2-8-docs.html#variable:CMAKE_INSTALL_PREFIX |
13:36.11 | starseeker | am I missing something? |
13:36.58 | starseeker | I'm seeing CMAKE_INSTALL_PREFIX as the last entry in my gui here |
13:40.54 | CIA-53 | BRL-CAD: 03d_rossberg * r42953 10/brlcad/trunk/src/conv/asc/g2asc.c: |
13:40.54 | CIA-53 | BRL-CAD: B_resolution (now B_unused) seams to be unused |
13:40.54 | CIA-53 | BRL-CAD: removed it |
13:44.13 | d_rossberg | i'm removing all my cache, build etc. files ... |
13:44.50 | starseeker | is trying on Windows to see if CMAKE_INSTALL_PREFIX perhaps isn't coming up there |
13:49.31 | *** join/#brlcad mafm_ (~mafm@193.153.53.189) |
13:50.40 | CIA-53 | BRL-CAD: 03d_rossberg * r42954 10/brlcad/trunk/misc/win32-msvc/Dll/BrlcadCore.def: removed disappeared function |
13:54.04 | starseeker | CMAKE_INSTALL_PREFIX is visible in my Windows setup too |
13:57.58 | d_rossberg | now here too, it was probable a problem with an outdated configuration |
14:00.36 | starseeker | nods. Yeah, a lot has changed over the past week |
14:01.23 | ``Erik | brlcad: did you take crit offline? |
14:04.09 | CIA-53 | BRL-CAD: 03starseeker * r42955 10/brlcad/trunk/src/conv/asc/g2asc.c: fix formatting string |
14:10.21 | CIA-53 | BRL-CAD: 03erikgreenwald * r42956 10/brlcad/trunk/TODO: The 'freebsd' (actually NDEBUG) strict issues are fixed. The metaball "shelling" bug is, as well. |
14:35.17 | CIA-53 | BRL-CAD: 03starseeker * r42957 10/brlcad/branches/cmake/misc/CMake/BRLCAD_Util.cmake: Don't assume -Wno-error will always work, check for it first |
14:57.31 | CIA-53 | BRL-CAD: 03starseeker * r42958 10/brlcad/trunk/src/liboptical/sh_toyota.c: Hmm - gamma shadows something in math.h on OSX 10.5, apparently... |
15:00.23 | CIA-53 | BRL-CAD: 03starseeker * r42959 10/brlcad/trunk/src/librt/primitives/brep/brep.cpp: These ARE used, at least on Windows... |
15:04.54 | CIA-53 | BRL-CAD: 03starseeker * r42960 10/brlcad/branches/cmake/ (18 files in 11 dirs): MFC r42959 |
15:35.45 | CIA-53 | BRL-CAD: 03starseeker * r42961 10/brlcad/branches/cmake/src/util/CMakeLists.txt: Can't do libpc on MSVC yet, so can't very well test it... |
15:44.05 | d_rossberg | starseeker * r42955: ups :{ |
15:44.21 | starseeker | np, happens |
15:53.10 | brlcad | ``Erik: nope, they did |
15:53.19 | brlcad | new server is now online |
15:54.07 | brlcad | mounting the drive and working on restoring passwd |
15:54.14 | ``Erik | aight |
15:54.27 | brlcad | new IP |
15:59.54 | ``Erik | going to give it a new name and remove the old one, or update? |
16:08.04 | CIA-53 | BRL-CAD: 03bob1961 * r42962 10/brlcad/trunk/src/tclscripts/archer/ (Archer.tcl ArcherCore.tcl): Remember whether or not to show things like the view axes, model axes, view params, scale etc. |
16:10.46 | brlcad | yes :) |
16:11.15 | brlcad | just got the drives mounted |
16:11.25 | brlcad | letting /home copy now |
16:26.05 | *** join/#brlcad yukonbob_ (~bch@20-144.wireless.kamloops.net) |
16:55.42 | CIA-53 | BRL-CAD: 03brlcad * r42963 10/brlcad/trunk/src/librt/primitives/ars/ars.c: this brings the ARS online with automatic v4 detection, though not a final solution since htons() will do nothing on other endian. |
18:11.59 | CIA-53 | BRL-CAD: 03starseeker * r42964 10/brlcad/branches/cmake/misc/CMake/FindTCL.cmake: Do a better job of generalizing the exec search options |
18:14.33 | CIA-53 | BRL-CAD: 03starseeker * r42965 10/brlcad/branches/cmake/ (misc/CMake/FindTCL.cmake src/other/togl/CMake/FindTCL.cmake): ws, update togl copy of FindTCL.cmake |
18:17.04 | ``Erik | starseeker: that did the trick |
18:17.21 | starseeker | cool |
18:17.26 | starseeker | is building now |
18:32.07 | ``Erik | hm, more failures on fbsd, linux, and osX.6 |
18:32.23 | ``Erik | 'src/other/tcl/unix/tclUnixPort.h:91:5: error: "TIME_WITH_SYS_TIME" is not defined' seems common |
18:32.38 | starseeker | CMake build or autotools? |
18:32.51 | ``Erik | cmkae |
18:48.19 | ``Erik | weeee, osX.6 GL.h has // style comments, so ogl fb craps itself with strict flags |
18:48.28 | ``Erik | /System/Library/Frameworks/OpenGL.framework/Headers/gl.h:37:1: error: C++ style comments are not allowed in ISO C90 |
19:06.20 | CIA-53 | BRL-CAD: 03brlcad * r42966 10/brlcad/trunk/src/libbu/xdr.c: group pairings together. makes it more obvious that bu_glonglong() was never implemented (though there is a macro) but doesn't matter since this is all getting deprecated. |
19:06.40 | CIA-53 | BRL-CAD: 03starseeker * r42967 10/brlcad/branches/cmake/misc/CMake/FindGL.cmake: We're not doing aqua based GL in these searches, so let's stay out of the opt/graphics directory and try /usr/X11 |
19:10.16 | CIA-53 | BRL-CAD: 03brlcad * r42968 10/brlcad/trunk/ (doc/deprecation.txt include/bu.h): deprecate the old eXternal Data Representation (xdr) API. the byteorder libc functions became IEEE Std POSIX.1-200x (with the exception of the long long pairing?) so set up the migration path. |
19:14.14 | CIA-53 | BRL-CAD: 03brlcad * r42969 10/brlcad/trunk/src/librt/primitives/ars/ars.c: note the problem |
19:15.38 | brlcad | ``Erik: yeah, that sucks |
19:16.11 | brlcad | where are you seeing it? |
19:16.29 | brlcad | I updated all of the places that gl.h gets included with ${STD_C99_FLAGS} |
19:16.57 | brlcad | that should squash that error |
19:17.10 | brlcad | libfb, libdm, and I think mged already have it |
19:20.31 | brlcad | ahh, if you're testing cmake build -- starseeker, you'll have to add similar logic |
19:21.32 | brlcad | you could actually just set c99 project-wide for cmake build since I'm close to doing that for autotools -- was waiting to squash the few remaining warnings after this release |
19:24.01 | ``Erik | osX.6 using cmake, grabbing OpenGL.framework instead of /usr/X11R6/include/GL |
19:24.58 | ``Erik | always amusing hearing starseeker use words you never think he'd use :> |
19:28.25 | CIA-53 | BRL-CAD: 03starseeker * r42970 10/brlcad/branches/cmake/misc/CMake/FindGL.cmake: Try the CMAKE_FIND_FRAMEWORK variable in FindGL |
19:47.17 | ``Erik | brlcad: using htons in ars.c will require winderz to link winsock.dll to librt |
19:47.41 | ``Erik | and include Winsock2.h apparently |
20:05.45 | CIA-53 | BRL-CAD: 03starseeker * r42971 10/brlcad/branches/cmake/misc/CMake/FindTCL.cmake: Need to reset variables if results aren't valid |
20:08.50 | ``Erik | (and arpa/inet.h on *nix) |
20:26.52 | starseeker | ``Erik: here's why it's crashing: 0x00000001004ea2bb in _X11TransWritev () |
20:29.09 | starseeker | ah, there's no escape - /usr/X11R6/include/GL/gl.h includes the framework gl.h |
20:31.06 | starseeker | that's trying to display mac to mac |
20:31.20 | starseeker | mac to Linux, wish itself gives up with a bad atom |
20:40.35 | brlcad | ``Erik: that htons is not going to stay there |
20:40.59 | brlcad | working on replacing that with something that'll actually work on big endian systems |
20:44.22 | CIA-53 | BRL-CAD: 03brlcad * r42972 10/brlcad/trunk/src/librt/primitives/ars/ars.c: switch over to the xdr functions while the FIXME is being fixed so that we don't have windows build fail. don't want to hook up winsock for windows. |
21:14.08 | starseeker | ``Erik: odd... runs on Windows for me here (VC++2010) |
21:32.50 | brlcad | ``Erik: loads of reallocated sectors on the old filesystem .. apparently a bad or failing drive? |
21:32.58 | brlcad | they're attempting a clone/recovery now |
21:33.38 | brlcad | it stalled hard 22GB into the restore earlier today and had to go diagnosing |
21:43.07 | ``Erik | old fs? on crit or bz? bz was falling apart hardcore |
21:43.08 | ``Erik | is |
21:43.17 | CIA-53 | BRL-CAD: 03erikgreenwald * r42973 10/brlcad/branches/cmake/src/other/tcl/ (5 files in 2 dirs): regex.h -> regextcl.h to avoid fighting with /usr/include/regex.h |
22:09.59 | CIA-53 | BRL-CAD: 03brlcad * r42974 10/brlcad/trunk/src/libbu/convert.c: |
22:09.59 | CIA-53 | BRL-CAD: address FIXME, make cv() be public API, so rename to bu_cv(). provides a |
22:09.59 | CIA-53 | BRL-CAD: simplified API means for converting between two specified formats, e.g. htons() |
22:09.59 | CIA-53 | BRL-CAD: would be something like bu_cv(outbuf, "nss", sizeof(short), val, "hss", 1); |
22:10.17 | CIA-53 | BRL-CAD: 03brlcad * r42975 10/brlcad/trunk/include/bu.h: expose the new bu_cv() function. |
22:15.30 | *** join/#brlcad Ralith (~ralith@S010600221561996a.vc.shawcable.net) |
22:27.33 | CIA-53 | BRL-CAD: 03brlcad * r42976 10/brlcad/trunk/src/libbu/convert.c: no longer vert.c |
22:39.42 | CIA-53 | BRL-CAD: 03starseeker * r42977 10/brlcad/branches/cmake/src/other/togl/src/CMakeLists.txt: Install the headers as well |
22:45.30 | *** join/#brlcad mafm_ (~mafm@203.Red-88-22-160.staticIP.rima-tde.net) |
23:04.38 | CIA-53 | BRL-CAD: 03brlcad * r42978 10/brlcad/trunk/src/librt/db_float.c: implement a simple flip_short() routine for flipping the bytes of a short. htons() isn't suitable because it won't flip bytes if host endain is network (i.e. big endian). this is private API. |
23:05.44 | CIA-53 | BRL-CAD: 03brlcad * r42979 10/brlcad/trunk/src/librt/ (Makefile.am librt_private.h): add a new private header, librt_private.h, so we can have a common header for providing functions that can be used within librt but that shouldn't be public API. |
23:09.04 | CIA-53 | BRL-CAD: 03brlcad * r42980 10/brlcad/trunk/src/librt/db_float.c: prepare for a rename to db_flip.c |
23:10.50 | CIA-53 | BRL-CAD: 03brlcad * r42981 10/brlcad/trunk/ (5 files in 2 dirs): rename from db_float.c to db_flip.c since the file is being expanded with a few more flip functions that have nothing to do with floats |
23:16.41 | *** join/#brlcad PrezKennedy (MK@whitecalf.net) |
23:25.34 | CIA-53 | BRL-CAD: 03starseeker * r42982 10/brlcad/branches/cmake/src/other/togl/ (12 files in 5 dirs): Let's see if glew can handle some of the cross platform annoyances |
23:28.51 | CIA-53 | BRL-CAD: 03starseeker * r42983 10/brlcad/branches/cmake/CMakeLists.txt: Turn Tk X11 requirement on if appropriate |
23:41.43 | starseeker | ``Erik: hmm: http://www.geodynamics.org/pipermail/cig-commits/2010-March/011594.html |
23:52.19 | starseeker | ``Erik: here's a backtrace, if that helps: http://pastebin.mozilla.org/1020171 |
23:57.06 | CIA-53 | BRL-CAD: 03brlcad * r42984 10/brlcad/trunk/src/librt/primitives/ars/ars.c: |
23:57.06 | CIA-53 | BRL-CAD: this should fix interpretation for both big and little endian platforms. if |
23:57.06 | CIA-53 | BRL-CAD: flipping is requested, we need to flip the bytes regardless of host/network |
23:57.06 | CIA-53 | BRL-CAD: endianness. call the new private flip_short() function from db_flip.c |