IRC log for #brlcad on 20080409

00:29.46*** join/#brlcad Twingy (n=justin@74.92.144.217)
00:32.10*** join/#brlcad Twingy (n=justin@74.92.144.217)
01:42.52*** join/#brlcad andrecastelo (n=chatzill@189.71.7.20)
01:43.17andrecastelogood evening
01:43.18andrecastelohi brlcad
01:54.03brlcadhowdy andrecastelo
02:05.04andrecastelogood night guys hehehe
02:05.16andrecastelotomorrow i'll be here the whole day :b hehe
02:05.18andrecastelocya
02:34.41starseekeryukonbob:  Yes and yes :-)
03:34.00yukonbobstarseeker: heh :)
04:41.18*** join/#brlcad pacman87 (i=127@resnet-45-192.dorm.utexas.edu)
04:44.31CIA-20BRL-CAD: 03starseeker * r30651 10/brlcad/trunk/src/proc-db/tire.c: Fix subtractions for ell intersections so wireframe does not extend so far beyond actual tire surface.
05:28.46*** join/#brlcad illethal (n=oden@c-71-200-221-159.hsd1.fl.comcast.net)
05:29.39*** part/#brlcad illethal (n=oden@c-71-200-221-159.hsd1.fl.comcast.net)
05:41.47*** join/#brlcad clock_ (n=clock@77-56-95-244.dclient.hispeed.ch)
07:03.37*** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch)
07:21.55*** join/#brlcad d_rossberg (n=rossberg@bz.bzflag.bz)
07:43.50brlcadyay brep csg reply
07:44.06brlcadwaves g'morning/g'night to d_rossberg
08:35.23d_rossbergwishes a well sleeping to brlcad!
09:56.23*** join/#brlcad mafm (n=mafm@elnet-111.lip.pt)
09:56.58mafmhallo
10:50.36*** join/#brlcad SJ (n=surya@203.200.95.130)
10:50.44SJhey guys
11:00.24mafmhey
11:01.52SJthis is hippieindamakin8 btw :)
11:06.00mafm:D
11:18.49SJsean can u gimme a link to Mr.Mike Muuss's papers on ray tracing
13:07.56*** join/#brlcad Axman6 (n=Axman6@pdpc/supporter/student/Axman6)
13:11.21*** join/#brlcad andrecastelo (n=chatzill@189.71.41.1)
13:12.19andrecastelohi everyone
13:13.47mafmhi
14:40.58brlcadyawns
15:00.27prasad_arrr
15:06.48``Erikyaaarrrgggghhhhhh, matey
15:08.45pooliobrlcad: shouldn't you bet at work already? :)
15:17.07mafmnobody should bet at work, specially with the bosses
15:19.32pooliomafm: yeah yeah...I'm still getting used to this keyboard
15:24.09mafmwhich keyboard is that?
15:27.56brlcadpoolio: yep
15:37.51*** join/#brlcad Elperion (n=Bary@p54873B47.dip.t-dialin.net)
15:42.20*** join/#brlcad andre|away (n=chatzill@189.71.41.1)
15:45.53*** part/#brlcad prasad_ (n=psilva@static-70-108-244-218.res.east.verizon.net)
16:37.25*** join/#brlcad andrecastelo (n=chatzill@189.71.41.1)
16:38.51yukonbobmorning cadheads
16:39.43andrecasteloyukonbob: good morning / afternoon :)
16:59.17andrecastelois there a point list structure already implemented in brlcad ?
17:01.57*** join/#brlcad prasad_ (n=psilva@static-70-108-244-218.res.east.verizon.net)
17:27.17CIA-20BRL-CAD: 03bob1961 * r30652 10/brlcad/trunk/misc/nsis/brlcad.nsi: ReadEnvStr must be called within a function or a section. Also, PROGRAMFILES cannot be set (i.e. its a constant defined by nsis). Put the ReadEnvStr inside the .onInit callback function.
18:01.19*** join/#brlcad clock_ (n=clock@77-56-82-18.dclient.hispeed.ch)
18:37.46*** join/#brlcad micges (n=michu@public-gprs66780.centertel.pl)
18:38.19*** part/#brlcad micges (n=michu@public-gprs66780.centertel.pl)
18:40.26mafmheading off, take care
18:51.47*** join/#brlcad docelic (n=docelic@78.134.198.227)
19:19.34*** join/#brlcad fi-dschi (n=philipp@p5498A033.dip0.t-ipconnect.de)
19:53.58fi-dschihello, is there somebody listening?
19:54.53fi-dschiI need some help as I don't succeed in compiling brlcad for arch linux i686
19:56.01fi-dschiYou may look at the last 263 lines of output at http://pastebin.com/m55261921
19:57.16fi-dschiIt seems to be a problem connected to asc2g
19:57.49fi-dschiMy CFLAGS and CXXFLAGS are -march=athlon-xp -O2 -pipe
20:48.21brlcadfi-dschi: hello
20:48.49fi-dschihello
20:49.09brlcadfi-dschi: I can't get to pastebin.com from current location, could you repost to pastebin.bzflag.bz ?
20:49.16fi-dschiok
20:50.43fi-dschibrlcad:http://pastebin.bzflag.bz/m42158c73
20:51.27prasad_who has vmware expreience
20:51.34brlcadfi-dschi: well the good news is that the build completed
20:51.57brlcad~translate de en Ungültiger Maschinenbefehl
20:52.26fi-dschiin en it is invalid machine code
20:52.44prasad_~translate jp en konichiwa
20:52.49fi-dschihello
20:52.51prasad_boo
20:53.02prasad_~translate en jp hello
20:53.10prasad_ah nice :)
20:53.55brlcadfi-dschi: if you run "src/conv/asc2g", what does it report?
20:54.00brlcadsame error?
20:54.03brlcador a usage statement?
20:54.52fi-dschitells me how to use
20:55.02fi-dschiUsage: asc2g file.asc file.g ....
20:55.20brlcadtry: cd db && ../src/conv/asc2g operators.asc operators.g
20:55.38fi-dschifrom inside src/conv?
20:55.44fi-dschiah, ok
20:56.02brlcadno, from inside db/
20:56.09fi-dschithis gives me "invalid machine instruction"
20:56.12brlcadrather .. from the top level
20:56.13brlcadhm
20:56.43brlcadthen from db/ try this:  ../libtool --mode=execute gdb --args ../src/conv/asc2g operators.asc operators.g
20:57.39fi-dschireports that there is no lobtool in ../
20:58.16brlcadhopefully no libtool?
20:58.24fi-dschiyes, of course
20:58.24brlcadnot lobtool :)
20:58.27brlcadok
20:58.53brlcadthat's odd .. are you building out of dir?
20:59.16fi-dschino, this is done by some tools
20:59.40brlcadfind /home/philipp/Software/abs/brlcad/src/brlcad-7.12.0 -name libtool
20:59.40fi-dschibut the files should be from the src archive
21:00.08fi-dschiok, it is therr
21:00.09brlcadhello andrecastelo :)
21:00.15fi-dschithis was my mistake
21:00.18brlcadthen ../libtool should have worked.. ;)
21:00.22fi-dschino
21:00.41fi-dschiah ok
21:00.42brlcad"then from db/ try this"
21:00.49fi-dschii didn't understand something else
21:00.52brlcadk
21:00.55fi-dschi../libtool: line 6566: exec: gdb: not found
21:01.02brlcadeep
21:01.03fi-dschithat's the output
21:01.05brlcadno gdb installed?
21:01.24fi-dschithanks
21:01.35fi-dschithis is missing as a deendency
21:01.55fi-dschiok, so I wil try again with gdb, thank you very much
21:02.10fi-dschithis will need some time
21:02.53brlcadwhat platform type is this?
21:03.03fi-dschiplatform type?
21:03.06fi-dschii686?
21:03.06brlcadhardware
21:03.12fi-dschiathlon xp
21:04.42brlcadhmm, an easy/quick test would be to take off the -march=athlon-xp option and see if the instruction error goes away
21:05.01fi-dschidoes it compile faster then?
21:05.12prasad_heh
21:05.25brlcadif you want a fast test compile, use the default options
21:05.38brlcadmaybe just ./configure --enable-all
21:05.43brlcadno flags
21:06.25fi-dschitis is all done by scripts, I normally type "makepkg" which will download the sources, compile and create a package for the pkgmanager of archlinux
21:06.41fi-dschibut I will try manually
21:11.23brlcadahh, the archlinux guy hangs out here
21:11.38fi-dschiyes, thats me
21:11.44brlcadoh heh
21:11.58fi-dschidid you read the package comments?
21:12.25brlcadare you louipc ?
21:12.31fi-dschino, this is the maintainer
21:12.34fi-dschii am fi-dschi
21:27.34brlcadokay, fi-dschi -- sorry about that :)
21:27.57brlcadfrom our perspective, we've only dealt with lou to date
21:28.13brlcadhe commits the archlinux changes directly to our repository misc/archlinux in your sources
21:31.55fi-dschihe told me to ask for help in this channel
21:39.05brlcadi highly suspect the problem has to do with the compilation settings and math computations, but it's really hard to say without seeing a stack trace
21:39.15brlcadthat error just usually indicates bad compiler ju ju
21:40.15fi-dschiis there still a problem? I thought there was simply gdb missing?
21:40.48brlcadhm?
21:41.34brlcadgdb was just to try to figure out why it was reporting a bad machine instruction
21:42.28fi-dschioh, so it should now give a more detailed error message?
21:45.31brlcadright
21:45.36brlcadto see if it was code or compiler
21:45.39brlcador both
21:46.01fi-dschiok, compilation has just finished
21:46.10fi-dschiand there are no errors as far as i can see
21:46.35brlcadah, cool
21:46.54brlcadthat would 'likely' be a compiler bug for that -march
21:47.09brlcador something being provoked with non-ieee floats or something similar
21:47.29fi-dschiok, i will try again with my cflags or what else can i doß
21:47.33fi-dschi?
21:47.53brlcadtry again with --enable-optimized later
21:48.23brlcadthat turns on a lot more optimization flags, still hardware generic, but should do much better than O2
21:48.50*** join/#brlcad louipc (n=louipc@bas8-toronto63-1096669639.dsl.bell.ca)
21:48.53brlcadhave you seen the PKGBUILD file loui wrote up?
21:48.56brlcadspeak of the devil
21:48.58louipc:D
21:49.01fi-dschi:D
21:49.07fi-dschiyes
21:49.08louipcyeah I was coming in for that exact purpose
21:49.17brlcadhowdy louipc
21:49.28fi-dschidid you get the post?
21:49.42louipcyep
21:51.13louipcyeah I took out any special configure flags, only --prefix is left because I had a problem with asc2g recently and was going to debug. but it worked fine with the flags removed hehe
21:52.55louipcbrlcad: greetings indeed
21:54.18louipcfi-dschi: maybe the ./configure output might give a clue... particularly the summary
21:54.49fi-dschifor the default ./configure? one moment
21:55.16louipcooh nevermind you found the problem?
21:55.36fi-dschime? no. i donÄt even understand what is going on!
21:55.56brlcadlouipc: the error is a run-time machine code error .. which is usually compilation settings or floating point data
21:55.57louipcdoes it work if you have gdb installed?
21:56.16fi-dschiwith ./configure --all-enabled, yes
21:56.43louipcbrlcad: yeah I was thinking something along those lines
21:56.49brlcadit worked if he removed his custom CFLAGS, which includes the -march optimization flag for a particular generation of athlons
21:57.20brlcadI suspect it's just that -march flag causing the problem
21:58.37louipcah
21:58.39louipcnods
21:59.08brlcadathlon-xp is either buggy or not the right architecture
21:59.16louipcfi-dschi: it should work with just generic x86_64 flags
21:59.44fi-dschiis this used by ./configure --all-enabled?
21:59.46louipcwell another user told me it works :D
21:59.56brlcad--enable-optimized should do the trick if performance is what you're going for
22:00.16brlcadfi-dschi: run "uname -a"
22:00.26fi-dschibut athlon-xp should be the optimized flag for me?
22:00.27louipcfi-dschi: ./configure by itself
22:00.43fi-dschiLinux PHPC_ARCH 2.6.24-ARCH #1 SMP PREEMPT Sun Mar 30 11:40:06 CEST 2008 i686 AMD Athlon(tm) AuthenticAMD GNU/Linux
22:00.55louipc:D
22:01.27fi-dschisome problems in the past made me think that athlon-xp was wrong
22:01.39louipcya
22:01.46fi-dschibut my cpu is called athlon-xp
22:01.55fi-dschiit is even printed on it
22:02.16louipcoh you're running i686 cool
22:02.47brlcad./configure --enable-all --enable-optimized CFLAGS="-mtune=i686"
22:02.52fi-dschithis is the archlinux default isn't it?
22:03.40louipcarchlinux default flags are CFLAGS="-march=i686 -mtune=generic -O2 -pipe"
22:03.48louipcif you go with that it should work
22:03.50fi-dschiok, i see
22:03.52brlcadthat doesn't necessarily mean that gcc thinks you have what it calls athlon-xp nor does it mean that the optimization loop in gcc isn't buggy ;)
22:04.43fi-dschibut shouldn't i be looking for the best fit?
22:04.43brlcadnotes that --enable-optimized is more agressive than "-march=i686 -mtune=generic -O2 -pipe"
22:04.49fi-dschiathlon, athlon-xp, i686?
22:05.14fi-dschiit has a higher priority?
22:05.20brlcadfi-dschi: it really won't matter -- the difference is going to be minimal
22:06.29fi-dschithere are just so many sites telling you which cflags you should use :)
22:06.40louipchehe
22:08.06fi-dschianother question to the above: what does "enable-optimized" do? does it e.g. use athlon-xp if this is the best fitting?
22:08.44louipcit takes twice as long ;D
22:09.07brlcadfi-dschi: in the summary at the end of configure, it tells ou the flags it'll use
22:09.14brlcads/ou/you/
22:10.50brlcadit basically turns on several optimization flags, almost as many as it can without modifying ieee behavior or becoming platform/hardware dependent
22:10.50fi-dschiso what are the settings in makepkg.conf for if the configure script can find the best flags on its own?
22:11.31fi-dschidoes this create platform depedency?
22:11.37louipcit's for packages which don't have such cool configure options :D
22:12.03brlcadarchlinux users may also just want to override the flags with their own, or make their build platform dependent (for that extra 2% performance)
22:12.31louipcthat too, but we're generally not ricers
22:12.32fi-dschiok, so let's set the flags back to default :)
22:12.44louipcfi-dschi: good idea
22:13.49louipcCXXFLAGS are the same as well
22:14.39fi-dschiyes, i know that
22:19.04fi-dschii will try to recompile with makepkg tomorrow. If there are problems you will hear from me again :). Thank ou so far and good night.
22:19.17fi-dschis/ou/you/ ;)
22:19.23brlcadcheers
22:19.25louipcalright take care
22:45.49*** join/#brlcad spike_ (i=[U2FsdGV@centaur.acm.jhu.edu)
22:46.19spike_is running off to the gym, brb
23:06.53brlcadcya

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.