IRC log for #brlcad on 20071002

00:38.54CIA-4BRL-CAD: 03brlcad * 10brlcad/src/tclscripts/archer/LoadArcherLibs.tcl: blt is actually already loaded as part of the preload in the main archer script, so don't need this load here
00:40.13CIA-4BRL-CAD: 03brlcad * 10brlcad/src/other/blt/src/bltInit.c: the minimum version checks are bogus for the pre-release alpha distribution, it needs to check all the way down to the patch level
00:41.17*** join/#brlcad tarzeau (i=sengun@berlin.ethz.ch) [NETSPLIT VICTIM]
00:41.17*** join/#brlcad louipc (n=louipc@bas8-toronto63-1096667620.dsl.bell.ca) [NETSPLIT VICTIM]
00:41.17*** join/#brlcad starseeker (n=CY@ip72-218-16-62.hr.hr.cox.net) [NETSPLIT VICTIM]
00:41.17*** join/#brlcad archivist (n=archivis@host217-35-76-52.in-addr.btopenworld.com) [NETSPLIT VICTIM]
00:41.19*** join/#brlcad tarzeau (i=sengun@berlin.ethz.ch)
00:42.07CIA-4BRL-CAD: 03brlcad * 10brlcad/src/archer/archer: additional try/catch checking for archer so that we can print better abort messages when things go horribly wrong
00:42.43*** join/#brlcad archivist (n=archivis@host217-35-76-52.in-addr.btopenworld.com)
05:42.26*** join/#brlcad Z80-Boy (i=clock@217-162-111-65.dclient.hispeed.ch)
06:04.04brlcadg'damn finally, archer now actually works on os x...
06:35.48CIA-4BRL-CAD: 03brlcad * 10brlcad/src/archer/archer:
06:35.48CIA-4BRL-CAD: the mighty bob lieth not. blt_library really _does_ need to be set AFTER blt is
06:35.48CIA-4BRL-CAD: loaded. the variable is wiped out during Blt_Init() yet is subsequently
06:35.49CIA-4BRL-CAD: required/expected in the Tabnotebook. this is the final mod that actually makes
06:35.49CIA-4BRL-CAD: Archer work on Mac OS X (tested, and seemingly fully functional) and probably on
06:35.51CIA-4BRL-CAD: Linux and BSD too now.
06:37.32CIA-4BRL-CAD: 03brlcad * 10brlcad/TODO:
06:37.32CIA-4BRL-CAD: archer is now functional, at least seems to be fully functional now on Mac OS X
06:37.32CIA-4BRL-CAD: with the slew of resource and library mods that were needed. archer runs both
06:37.32CIA-4BRL-CAD: installed an even uninstalled from a compiled source directory.
06:44.38CIA-4BRL-CAD: 03brlcad * 10brlcad/TODO: dupe
06:49.24yukonbobbrlcad: nice!
06:51.52CIA-4BRL-CAD: 03brlcad * 10brlcad/ (7 files in 2 dirs):
06:51.53CIA-4BRL-CAD: already deprecated since 7.0, consider the following files finally obsolete for
06:51.53CIA-4BRL-CAD: the next minor update: msr.h, rtstring.h, rtlist.h, shortvect.h, shortvect-pr.h;
06:51.53CIA-4BRL-CAD: the contents of those files moved into or were replace by functionality in the
06:51.53CIA-4BRL-CAD: bu.h header
06:54.28CIA-4BRL-CAD: 03brlcad * 10brlcad/TODO: there are a variety of install and post-install support scripts now, aside from being an overly vague/generic todo item
07:20.52*** join/#brlcad elite01 (n=elite01@dslb-088-070-005-186.pools.arcor-ip.net)
07:30.31*** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch)
07:35.43CIA-4BRL-CAD: 03brlcad * 10brlcad/ (TODO doc/deprecation.txt): move the deprecation items to their own file so that they can actually be preserved in a running log of API changes, including basic guidelines for how to go from deprecated to obsolete.
07:45.08CIA-4BRL-CAD: 03brlcad * 10brlcad/TODO: implement a lighting model for rt that visualizes how expensive each pixel is to compute via some heat graph.
07:45.26*** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch)
10:31.55*** join/#brlcad elite01 (n=elite01@dslb-088-070-005-186.pools.arcor-ip.net)
10:57.58*** join/#brlcad elite01_ (n=elite01@dslb-088-070-005-186.pools.arcor-ip.net)
12:52.09*** join/#brlcad elite01_ (n=elite01@dslb-088-070-020-061.pools.arcor-ip.net)
13:34.59*** join/#brlcad thing0 (n=ric@i157-115.nv.iinet.net.au)
13:52.46L0CALZ_0NLYbrlcad: what about taking rtedge and in the black pixels and neighbouring ones raytrace it with higher resolution than in the remaining ones?
13:53.06L0CALZ_0NLYit should work fine on models without texture mapping
14:16.47yukonbob...on to tclcadAutoPath.c.
14:41.29``ErikO.o
15:15.27*** part/#brlcad thing0 (n=ric@i157-115.nv.iinet.net.au)
15:24.31CIA-4BRL-CAD: 03erikgreenwald * 10brlcad/bench/run.sh:
15:24.31CIA-4BRL-CAD: Add another path for RT.
15:24.31CIA-4BRL-CAD: Search for 'moss.g' in $DB and cull (so if the .g's are in a different dir than $(top_srcdir)/db, the right path is used).
15:43.27*** join/#brlcad thing0 (n=ric@i157-115.nv.iinet.net.au)
15:46.12*** join/#brlcad cad28 (n=89f4d713@bz.bzflag.bz)
16:01.50*** part/#brlcad thing0 (n=ric@i157-115.nv.iinet.net.au)
16:16.02*** join/#brlcad poolio (n=poolio@c-71-206-215-46.hsd1.pa.comcast.net)
16:19.44CIA-4BRL-CAD: 03erikgreenwald * 10brlcad/src/rt/view.c: Disable image restart.
16:25.07MinuteElectronbrlcad: sorry to bug but a bit of info about the current website status would be nice.
17:03.10*** join/#brlcad frbrgeorge (n=george@GW.itcomm.ru)
17:03.48frbrgeorgehi there
17:05.38frbrgeorgeI've got a problem with autogen/configure BRL-CAD
17:07.10``ErikO.o
17:07.52*** join/#brlcad AchiestDragon (n=david@whipy.demon.co.uk)
17:08.35frbrgeorgeusing -l<library> in LDFLAGS instead of LIBS in autogen.sh cause modern ld failed
17:10.03``Erikhum, where are you seeing that?
17:11.09frbrgeorgewe have --as-needed bu default in our distro
17:11.44frbrgeorge(I mean ld --as-needed)
17:12.19``Erikah, a debian or ubuntu?
17:12.46``Erikwith a broken libtool? heh
17:13.18frbrgeorgeso "cc conftest.c -lpng" works, but "cc -lpng conftest.c" doesn't
17:13.34frbrgeorgeno, ALT Linux ;)
17:14.12``Erikodd, so it's breaking at configure time? with the distributed configure script?
17:14.55``Erik<-- doens't have any -l stuff in any of his LDFLAGS vars, grepped it all
17:15.42``Erikmebbe if you could post the offending bit of config.log to, say, http://pastebin.bzflag.bz/ ?
17:15.49frbrgeorgeno it just can't foind libpng, libx11 etc, but not breaks :)
17:19.27frbrgeorgewell, let me check my unhack first...
17:28.54frbrgeorge...it partly works
17:33.29frbrgeorgethere's nothing special to paste, but http://pastebin.bzflag.bz/pastebin.php?dl=d8910551
17:33.54*** join/#brlcad qazwsx (n=qazwsx@unaffiliated/adgj)
17:34.44qazwsxI've been playing with mged and I was wondefing if there are operationsin brlcad like the following: " here's a line segment; rotate it about this axis; get me a curved surface"; "here's a line segment; extrude it in this direction, get me a surface", ... "here is a line segment; here is a sphere, run the sphere through the line, get me the corresponding surface"
17:35.35``Erikand when you flip where the -lpng is on that compile line, it works?
17:36.01frbrgeorge``Erik: yep
17:36.05``Erikqaz: not really, it's a solid modeller... the last one can be done with a pipe or cylinder and sphere's unioned on the ends...
17:36.21qazwsx``Erik: any suggestion of anther tool?
17:37.18``Erikfrbrgeorge: that's a straight up AC_CHECK_LIB() ... I dunno what to tell you with that :/
17:37.34``Erikqazwsx: you mean like qcad or blender? O.o
17:38.20frbrgeorge``Erik: is autogen.sh                                                            15,1           Top
17:38.37frbrgeorge``Erik: (oops) is autogen.sh a generated file
17:38.38frbrgeorge?
17:38.43``Erikautogen.sh should just call autoreconf with some flags
17:39.01``Erikit's a script that brlcad found or wrote and has been tweaked a bit
17:39.17frbrgeorge``Erik: (oops twice:) I mean configure.ac, neturally
17:40.05``Erikerm, no? O.o
17:44.14frbrgeorge``Erik: so if configure.ac is written manually, one has to replace those LDFLAGS="$LDFLAGS $LIBPNG" (etc.) clauses with LIBS="$LIBS $LIBPNG"
17:47.14``Erikahhhh, so you're catching it on the AC_TRY_RUN(), not the AC_CHECK_LIB() ?
17:50.08frbrgeorge``Erik: buth :(
17:50.36``Erikthere's nothing we cna do about the AC_CHECK_LIB(), that's before those variables get set
17:51.07frbrgeorge``Erik: no, AC_CHECK_LIB seems to work
17:51.53``Erik<-- testing a fix...
18:23.20CIA-4BRL-CAD: 03erikgreenwald * 10brlcad/configure.ac: LIBS instead of LDFLAGS to hold library names (thanks to frbrgeorge for spotting this one)
18:25.46MinuteElectron]/w 32
18:33.03yukonbob+++HANGUP
18:34.26``Erikheh, you mean +++ATH0 ?
18:34.36``Erikhayes codes, pheer
18:35.31yukonbobatdt [erik's l33t 'puter]
18:36.22``Erikheh
18:36.29``Erikatpt !
18:36.30``ErikO.o
18:36.45``Erikpopopopop popopopop popoppopopopop
18:37.12yukonbobah pt..
18:37.15yukonbobdoesn know.
18:37.19yukonbob*does
18:38.18yukonbobthose should be attp, attd
18:38.26``Erikoh, atp or atdp
18:39.02yukonbobah right ... what is atpt?
18:39.09yukonbobatdp == dial pulse.
18:39.14``Erikme misrecalling and thinking atp
18:39.25yukonbobpheer indeed ;)
18:39.31``Erikit's been a long time
18:39.37``Erikpoke 53280,0
18:39.38``ErikO.o
18:39.40yukonbobtogether we'll be able to get a c64 online
18:39.43``Erikc64 badassedness
18:39.46yukonbobhahah
18:39.51yukonbob6510 baby..
18:40.10yukonbobonly  ml I know...
18:40.31``Erikthe coleco adam was a neat machine, but the c64 was better at games
18:41.03``Erik<-- has done too many assembly languages and monitor/mnem :/
18:41.53``Erikdambusters? at night?
18:42.02``Erikso pretty much an all black screen O.o heh
18:42.13yukonbobdambusters sounds right.
18:43.24``Erikall in past tense, obviously this person is no longer your friend after dumping one of those things on 'em ;) *duck*
18:43.44``Erikhum, dragons lair, buck rogers, ...
18:43.52yukonbobheh -- person wasn't my friend, and writing a book on it was _their_ idea, not mine ;)
18:44.08``Erikhrm, what was the one with the helicoptor doing sorties to rescue people... with the three buildings...
18:44.40yukonbobfor coleco/adam, me prefered the donkey kong series
18:45.04yukonbobb.c. quest for tires
18:45.08yukonbobsmurfs
18:45.34``Erikheh, yeah
18:49.16``Erikheh, gorf
18:49.56``Erikzaxxon
18:50.40``Erikchoplifter !
18:50.48``Erikdamn, that was driving me nuts
18:50.52yukonbobthat rings a bell.
18:51.20``Erikit was significantly ported
19:19.12``Erikallllllrighty then, BSD4.3 on a vax 11/780
19:26.37brlcadscary
19:32.25yukonbobbrlcad: re: that X include yesterday -- was a pkgsrc peculiarity; nothing to do w/ the distribution or the flags.
19:33.05yukonbobnow I'm onto a tclcad issue (when I get some spare cycles)
19:43.43brlcadqazwsx's inquiry can actually be done for the second case as well -- that's just a sketch and an extrude; the surface of revolution and sweep are the ones we can't do, but linear extrudes are old news
19:45.35*** join/#brlcad Z80-Boy (i=clock@77-56-71-35.dclient.hispeed.ch)
19:46.28brlcadyukonbob: tclcad issue?
20:02.14*** join/#brlcad AchiestDragon (n=david@whipy.demon.co.uk)
20:02.33yukonbobbrlcad: http://www.pastebin.ca/723441
20:05.21brlcadick
20:05.35brlcadundoutedly the const modifier .. haven't seen that in a while
20:12.31brlcadiirc, there's a one-line fix to a tcl header that fixes the problem if it's the one i'm thinking of .. but what was that line...
20:38.46``Erikhrmmmmm, someone said something about long long a while ago and someone said we don't use 'em... that ain't true, tcl.h has 'em :(
20:39.27``Erikand that spew of errors almost looks like that regeh.h issue O.o
20:46.05``Erikhttp://pastebin.bzflag.bz/m1a1e1c3d
20:47.01yukonbob``Erik: =)
20:56.18``Erikdoh, that'll teach me to rely on a tmp dir, heh
22:08.39CIA-4BRL-CAD: 03brlcad * 10brlcad/src/tab/scriptsort.c: declare yyin
22:09.50CIA-4BRL-CAD: 03brlcad * 10brlcad/src/tab/Makefile.am: disable scriptsort if we don't have a suitable lexer/parser since it needs a lexer (fixes build on mjm)
22:11.01brlcad``Erik: awesome
22:11.25brlcadgetting the build to work will probably be a heck of a lot easier with --enable-only-benchmark
22:12.17brlcadhopefully you have an up-to-date gcc on there though.. I'd had to revert back to patches that're needed for old 4.3 compiler to work
22:23.46``Erikheh, no, it's the cc that comes with bsd, all encumbered... :/
22:24.09``ErikI was thinking about checking out an old old old version from the vgr era
22:24.37``Erikbut I'll have to talk to kermit to figure out exactly what kinda hw is involved... I'm told the machine was bought with a single processor and field upgraded to dual at some point
22:46.00brlcadyou know where the sources are right?
22:47.38brlcad/c/BRL-CAD/OLD on cad, you can get 3.0 or 4.0 from there .. 4.5 or 5.0 is probably the last version to be tested on vgr before it was decommissioned
22:53.30``ErikI was gonna just cvs 'em out
22:54.02``Erikpulling source from '84 has been... edjimucational :) far more readable I think
22:55.29brlcadthere was a lot less of it
22:55.52``Erikyes, less files, and they're much simpler and shorter
22:55.54brlcadtwo decades of feature creep and enhancing
22:56.02``Erikraytrace.h was especially useful :)
22:56.38brlcadthat was a fun exercise, finding all of those old versions and tagging them
22:56.56brlcadI didn't try to compile the really old ones, just tagged for reference
22:57.01``Erikheh
22:57.03brlcadsince some files had moved/renamed
22:57.13``Erikyeah, and early on, ther'es not even a build system
22:57.14brlcadand would have been mostly futile/incompatible
22:57.22brlcadyeah
22:57.29``Erikand some of the tagged ones don't even have all the headers referenced in the code
22:57.38brlcadfor a while, mike used to say he'd just cat all the files together and pipe it to the compiler
22:58.01``Erikheh, I suppose that's one way... provided your headers are correctly gaurded
22:58.32brlcadproperly guarded and/or only included once
22:59.10``Erikincluded once in a C file, perhaps, but if two files include, say, stdio.h ... and it's not properly gaurded *shrug*
22:59.24``ErikI tried to compile some of the earlier stuff, but the systems have changed too much
22:59.34brlcadwelp.. got a really preliminary gross rough estimate vgr for mjm
22:59.35``ErikI couldn't even get an early jove to compile, and that had a makefile
22:59.45``Erikall cores lit up?
22:59.53``Erikdid it break 100k?
23:00.13brlcadI haven't done the distribution yet
23:00.32``Erikbe interesting to render a HUGE image and walk the cpu count to get a scalability set
23:00.35brlcadif it scaled perfectly linearly, and presuming the node I used was even representative ...
23:00.55brlcadwhich of course it won't
23:00.57brlcadbut if it did
23:00.59``Erikfor count in `jot 5000` ; do ./ rt ...
23:01.00``Erik:)
23:01.04brlcadthe vgr would be
23:01.08brlcad*drumroll*
23:01.14``Erikget on with it
23:01.15brlcad8897900
23:01.20``Erikhum
23:01.32``ErikI'd still like to see the scalability graph
23:01.40brlcadyep, that's next
23:01.51brlcadif it hits half that, I'll be amazed
23:01.58``Erikat that many cores? hell yeah
23:02.08``Erikum, it's a cluster, so that'd be remrt?
23:02.12brlcadthough I'm talking with msrc now to see if I can schedule interactive on the entire cluster at some point
23:02.44``Erikthey're upset that their resources aren't be used as much as they think they should
23:02.44brlcadotherwise, I'm going to give the batch system a try
23:03.07brlcadthe new hawk system is freaking sweet
23:03.19brlcadthat 'should' hit top 25 in the top500
23:03.29``Erikhawk?
23:03.43brlcadnew 9000 processor altix
23:03.55``Erikah... wow...
23:04.04brlcad18x512 iirc
23:04.08``Erikthose are still being purchased?
23:04.31``Erikdoes intels cc do decent vliw optimizations yet?
23:04.38``Erikgcc's output on those suck
23:04.39brlcadafaik, or purchased and being installed already from the sounds of it
23:05.28brlcadlast I tested icc vs gcc on altix, it kicked gcc's ass, but only with -gen_prof/-gen_use
23:05.42``Erikheh
23:05.52``Erikbrute force finding of optimization paths? O.o
23:05.58brlcadthat was a couple years ago, and in all fairness didn't try gcc's dual pass opt
23:06.18``Erikuhhh, what's that compiler company twingy was working with? uh, f something...
23:06.35brlcadyeah, eliminate all the branchs by at least making all the first guesses correct
23:06.48brlcaddunno
23:07.31``Erikthey had a compiler out and their claim was that if any compiler produced faster code, it was a bug... they had a funky script that went through every permutation of optimization flags to achieve that, if I understood correctly
23:07.51``Erikand, uh, adrt/tie came out faster using gcc than their stuff hehehhe
23:08.02brlcadand remrt isn't particularly awesome, though the amount of data being transferred is exceptionally minimal (just the ray out and pixel results back), so it really doesn't do too bad a job
23:08.17``Erikhm
23:08.24brlcadit's more a pita to set up as it was never designed for ease of use -- it was designed to get some jobs crunched out some year
23:08.32brlcadwhilest making movies
23:08.33``Erik*nod*
23:08.53``Erikso every transaction is a single ray/pixel handshake?
23:09.09``Erikbundling up sets would probably help a lot
23:09.16brlcadno, it has batching
23:09.20``Erikokie
23:09.34``Erikbut tcp/ip?
23:10.01``Erikat some point, the latency and stack overhead will overshadow the cost of just doing it :/
23:10.18``Erikoh, I turned off the 're-use pix data' in rt...
23:10.33``Erikafter my bsd/vax adventure, I might try wiring the bu_image_save() stuff into rt
23:10.43``Erikbut vax is more fun :D
23:11.57``Erikshoot,I forgot to email dwayne to let him know where a clone enabled install is
23:12.58``Erikhttp://www.collegehumor.com/picture:6129   hehhe one of those doesn't belong... or... does it?
23:13.48``Erik(that page is work safe, btu most of the site is not)
23:13.58brlcadi saw the reuse bit.. should rip the code out and annotate news
23:14.11brlcadmake sure there arent other outliers
23:14.25brlcadrelated to automatic restart
23:15.11``Erikyeah... I did the #if 0 trick... wanna be slow about removing possibly used functionality :/
23:16.17brlcadthat's not exactly slow .. as good as ripping it out to the user :)
23:16.24brlcadi mean, it's in cvs
23:16.34brlcadif it need to be reverted, it can be on a dime
23:17.28``Erikmeh *shrug*
23:18.21brlcadplus next release is minor ver #, so good time for it
23:18.54brlcadotherwise, I'd think we should just leave it on with a bu_log'd deprecation message to see if anyone notices
23:19.41``Erikheh
23:20.27``ErikI'm doing an update to gut it... personally, I'd probably go with a bu_log() and have it disabled... or mebbe a bu_bomb() to get someone actually talking to us
23:25.41``Erikheh
23:25.47``Erikdomo arigato mr roboto
23:26.30``Erikah, unicode math symbols
23:27.46starseekerThey're sort of the "Duke Nukem Forever" of the scientific software world...
23:32.42``Erikhehehe
23:32.52``Erikread a webcomic recently about halo three, uhhhhhhh
23:32.59``Eriklemme find it
23:33.14louipc"Take me, Robert!" sighed the forty-five year old construction worker as the tripe-fondling construction worker diddled her moist ear lobes and augured his swinish gusher into her dusty appendix scar.
23:34.35``Erikhttp://godmodeonline.com/d/20070924.html
23:34.46Twingypathscale
23:39.04``Erikyeahhhhh, them, amusing stuff :) compilers are fun O.o
23:39.40``Eriktwingy, didja get that fpga chunking along? the development softwawre was all lame and I think they wanted you to buy a real license to really use it...
23:40.41``Eriknot to mention the windowsyness of it all
23:53.00CIA-4BRL-CAD: 03brlcad * 10brlcad/bench/Makefile.am: set the LD_LIBRARY_PATH and DYLD_LIBRARY_PATH so that we can find tcl/tk prior to installation

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