| 01:22.31 | CIA-4 | BRL-CAD: 03johnranderson * 10brlcad/src/mged/ (chgview.c cmd.c): |
| 01:22.31 | CIA-4 | BRL-CAD: "B -A" no longer acts like you gave it the wrong syntax if nothing matches. |
| 01:22.31 | CIA-4 | BRL-CAD: Also, "B" commanad with no args now acts like "Z". |
| 02:58.58 | ``Erik | 'commanad' wow... |
| 02:59.26 | ``Erik | yeah, he's just a dude, but, ... jra produces godlike output, it boggles me that he'd make a typo\ |
| 04:56.51 | brlcad | heh |
| 06:08.31 | *** join/#brlcad elite01 (n=elite01@195.37.106.60) |
| 07:16.13 | *** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch) |
| 07:24.54 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/src/other/ (4 files in 4 dirs): create/generate the generic versionless libraries during compile/install so that the former behavior of just using -ltcl and -ltk will work for external codes. |
| 07:39.31 | CIA-4 | BRL-CAD: 03brlcad 07STABLE * 10brlcad/src/other/ (4 files in 4 dirs): merge the tcl/tk symlink checks in from head so that -ltcl and -ltk work for external users |
| 09:16.56 | *** join/#brlcad elite01 (n=elite01@dslb-088-070-018-050.pools.arcor-ip.net) |
| 10:56.09 | *** join/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 12:52.00 | *** part/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 12:52.16 | *** join/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 12:58.08 | *** part/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 13:07.58 | *** join/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 13:10.28 | *** join/#brlcad elite01 (n=elite01@dslb-088-070-029-074.pools.arcor-ip.net) |
| 13:54.53 | ``Erik | heh |
| 13:55.21 | ``Erik | I'm out sick today. I woke up feeling like shit, then got an email telling me that a close college buddie died in a motorcycle accident. |
| 13:55.44 | Z80-Boy | ``Erik: was it like telepathy? |
| 13:55.49 | Z80-Boy | Feeling shit because someone close died? |
| 13:55.52 | ``Erik | no, he died a month ago |
| 13:56.05 | ``Erik | blew a corner and hit a sign :( |
| 13:56.25 | ``Erik | as much as I like motorcycles, I doubt I'll ever own one. |
| 13:57.04 | ``Erik | I've ridden them before, and have had fun... but hearing stories, losing friends... and the idiocity of suv drivers... it's not a very safe activiy |
| 13:58.00 | ``Erik | so, brlcad, I tried to call phb (got vm), and sent email, if anyone is looking for me, save it 'till tuesday :D |
| 14:12.14 | Z80-Boy | ``Erik: Mike Muus died on a road... |
| 14:28.46 | *** join/#brlcad MinuteElectron (n=MinuteEl@silentflame/member/pdpc.base.minuteelectron) |
| 14:36.41 | ``Erik | 7 years ago |
| 14:37.06 | ``Erik | but I (unfortunately) never met the man |
| 14:37.57 | ``Erik | so it's a BIT more distant than finding out a friend who I've written code with, went out drinking with, slept on his couch (after drinking too much)... :/ |
| 14:38.04 | ``Erik | a month ago |
| 14:38.28 | ``Erik | at least it was an accident... a close friend of mine commited suicide a couple years ago, THAT was hard |
| 14:39.47 | ``Erik | ended up bawling like a baby over that :) right now, I'm just drinking myself fucktarded |
| 14:41.03 | Z80-Boy | was it hard to commit the suicide? |
| 14:41.22 | Z80-Boy | Two classmates from the university, one was a friend, also comitted suicide |
| 14:41.50 | Z80-Boy | a distant friend let's say |
| 14:42.37 | ``Erik | no, my coping with it was hard. |
| 14:43.02 | Z80-Boy | like some programs corrupt their memory and then crash |
| 14:43.04 | Z80-Boy | people do the same |
| 14:43.21 | Z80-Boy | convince themself that jumping from a bridge etc. is good for them |
| 14:43.30 | Z80-Boy | ideas can be dangerous ;-) |
| 14:43.40 | ``Erik | rocky got me my first computer job, I sat in his living room for new years 2000, with the couch that had a broken spring that stabbed your ass, heh |
| 14:44.04 | Z80-Boy | no more rocky anymore because rocky crashed, that simple |
| 14:44.10 | ``Erik | he was a great guy :/ |
| 14:44.29 | ``Erik | dan was, too |
| 14:44.47 | ``Erik | in general, I will only get fitshaced drunk infront of people I really really trust. these were both guys that qualify for that |
| 14:44.51 | Z80-Boy | don't worry he would be dead later anyway |
| 14:44.58 | Z80-Boy | all people seem to eventually become dead |
| 14:45.07 | Z80-Boy | he just time shifted it |
| 14:45.10 | ``Erik | heh |
| 14:45.12 | Z80-Boy | like you time shift a TV show on VHS |
| 14:45.25 | ``Erik | when my grandparents passed, it was a lot easier |
| 14:46.19 | ``Erik | I'm not sure if it's the lost possibilities, or the realization of mortality, but young people dying is a lot more tragic than old people dying |
| 14:46.34 | ``Erik | vodka good. *grunt*. |
| 14:46.54 | Z80-Boy | well the young people would turn into old people anyway |
| 14:48.05 | Z80-Boy | and then into zombies |
| 14:48.09 | Z80-Boy | and then they would turn dead |
| 14:48.42 | Z80-Boy | dying should be banned under the death penalty |
| 14:48.59 | Z80-Boy | cause it makes the remaining people uncomfortabl |
| 14:49.10 | ``Erik | attempted suicide is a federal crime in the US |
| 14:49.24 | Z80-Boy | so your friend is a criminal! |
| 14:49.31 | ``Erik | no |
| 14:49.38 | Z80-Boy | can you get into jail for attempting suicide? |
| 14:49.38 | ``Erik | he didn't attempt, he succeeded. |
| 14:49.41 | ``Erik | yes |
| 14:49.45 | Z80-Boy | wow |
| 14:50.11 | Z80-Boy | murder on /dev/loopback |
| 14:50.31 | ``Erik | http://en.wikipedia.org/wiki/Legal_views_of_suicide |
| 14:50.37 | Z80-Boy | This is the police. Stop aiming at your head or we shoot you down |
| 14:50.57 | Z80-Boy | Ironically, the punishment for attempted suicide in some jurisdictions has been death. |
| 14:51.00 | Z80-Boy | lol |
| 14:51.02 | ``Erik | ok, not federal, some states |
| 14:52.30 | Z80-Boy | In the United States, suicide has never been punished as a crime nor penalized by property forfeiture or ignominious burial. |
| 14:53.37 | ``Erik | I'm more in tune with teh laws when I grew up, so I'm kinda operating on early 90's law... with washington being a specific set... which listed suicide as a felony at the time *shrug* |
| 14:53.54 | ``Erik | I have been educated. Suicide is no longer a punishable crime in the us |
| 14:56.38 | Z80-Boy | Well suceessfull suicide is still punished by the death penalty |
| 15:08.39 | *** join/#brlcad tarzeau (i=gurkan@bee.ethz.ch) |
| 15:08.41 | *** part/#brlcad tarzeau2 (i=sengun@berlin.ethz.ch) |
| 15:16.25 | *** join/#brlcad poolio_ (n=poolio@c-71-206-215-46.hsd1.pa.comcast.net) |
| 15:19.50 | *** join/#brlcad Elperion (n=Bary@p54875A64.dip.t-dialin.net) |
| 15:53.50 | starseeker | ``Erik: Sorry to hear that - my condolences :-( |
| 16:42.42 | *** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch) |
| 17:00.49 | brlcad | howdy yukonbob |
| 17:02.16 | yukonbob | hey brlcad. Happy Friday.. |
| 17:02.26 | brlcad | yeah, likewise :) |
| 17:05.13 | *** topic/#brlcad by brlcad -> BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || BRL-CAD is on scan.coverity.com, ask for an account if you're willing to hunt/fix bugs :) || Release 7.10.4 imminent |
| 17:05.19 | brlcad | ~spell imminent |
| 17:05.30 | brlcad | sounds too minty |
| 17:05.36 | Z80-Boy | ~spell immenent |
| 17:05.47 | Z80-Boy | ~spell eminem |
| 17:05.55 | brlcad | ~dict immanent |
| 17:06.02 | Z80-Boy | ~spell brlcad |
| 17:06.19 | brlcad | ~dict imminent |
| 17:06.33 | brlcad | aiight, so I did get it right at least :) |
| 17:06.46 | Z80-Boy | ~spell bricket |
| 17:07.30 | bricked_brigade | ~spell briquette |
| 17:08.23 | yukonbob | brlcad: I'll take an account for coverity -- I assume it's up/running again? |
| 17:08.51 | brlcad | it's up, but the scan is incomplete |
| 17:09.06 | brlcad | so it's not really of any use just quite yet until the scan is rerun |
| 17:09.19 | bracket_brigade | brlcad: I think you should render head and headcut from ronja to see the vast orders of magnitude rendering time |
| 17:09.28 | bracket_brigade | when one is just a cutaway view of the other |
| 17:09.46 | bracket_brigade | or perpend and perpendcut it's a simple design and it's the same problem |
| 17:10.39 | brlcad | have you run a profile? |
| 17:10.54 | brlcad | that would be useful to see if the difference is obvious |
| 17:10.59 | brlcad | i mean code-wise |
| 17:18.45 | brlcad | ahh .. apparently I can't create accounts just yet even if I want to |
| 17:18.59 | brlcad | david has to enable some option or create them for us |
| 17:19.05 | brlcad | probably because our scan was incomplete |
| 17:19.40 | *** topic/#brlcad by brlcad -> BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || BRL-CAD is on scan.coverity.com but not completely set up yet, more news TBD || Release 7.10.4 imminent |
| 17:19.46 | bracket_brigade | a profile? What is it? |
| 17:20.07 | bracket_brigade | Like code profiling in C? |
| 17:24.01 | brlcad | yeah |
| 17:24.02 | brlcad | gprof |
| 17:24.21 | brlcad | ./configure --enable-profiling --enable-debug --disable-optimized |
| 17:24.54 | brlcad | then run one ray-trace, run gprof; then run the other ray-trace, and run gprof again |
| 17:25.08 | bracket_brigade | gprof with what params? |
| 17:25.10 | brlcad | post/compare the two reports |
| 17:25.27 | brlcad | when you run rt it'll dump out a gmon.out file into the current directory |
| 17:25.46 | brlcad | so you just have to run gprof in that same directory, with a param of the /path/to/rt |
| 17:26.26 | brlcad | e.g. /usr/brlcad/bin/rt -o /dev/null ronja.g ronja && gprof /usr/brlcad/bin/rt |
| 17:26.40 | bracket_brigade | aha nice |
| 17:26.47 | brlcad | er, gprof /usr/brlcad/bin/rt > results.log 2>&1 |
| 17:27.20 | brlcad | then rm gmon.out, do another rt with the slow object, and get the next gprof output |
| 17:48.02 | *** join/#brlcad Z80-Boy (n=clock@zux221-122-143.adsl.green.ch) |
| 17:48.17 | Z80-Boy | If I run ./configure with different parameters do I have to type make clean after? |
| 17:48.24 | Z80-Boy | Or can I continue with make? |
| 17:51.31 | brlcad | have to make clean |
| 17:51.41 | brlcad | it has to rebuild all of the files with the profile parameter |
| 17:51.54 | Z80-Boy | gprof a.out no such file or directory |
| 17:52.23 | Z80-Boy | aha I didn't do make clean anyway |
| 17:52.52 | brlcad | you wouldn't have an a.out either |
| 17:52.57 | Z80-Boy | hmm gprof reads the given object file but I guess rt is assembled from a lot of different object files which one should I supply? |
| 17:52.58 | brlcad | unless you're building your own thing |
| 17:53.08 | Z80-Boy | The one whose timing I am interested in? |
| 17:53.33 | brlcad | you supply gprof with the binary that you're running (i.e. rt) |
| 17:53.42 | brlcad | one that ideally has profile symbols included |
| 17:53.44 | Z80-Boy | OK cool |
| 17:54.05 | brlcad | to get those profile symbols, you have to build the application (and all object files and all libraries it uses) with a profile option enabled |
| 17:54.21 | brlcad | otherwise it can't instrument the function calls and give a good profile |
| 17:54.51 | brlcad | so yeah, make clean, run configure with those options I mentioned, then run rt+gprof twice |
| 18:07.45 | CIA-4 | BRL-CAD: 03brlcad * 10brlcad/NEWS: john fixed mged 'B' command behavior with -A and no args |
| 18:08.41 | yukonbob | .and "B" alone doesn't puke error msgs anymore |
| 18:12.03 | brlcad | yep |
| 18:12.11 | brlcad | acts like Z |
| 18:12.56 | yukonbob | is it possible to have lookat behave like eye_pt or ae where if it has no args, it reports current settings? |
| 18:14.40 | brlcad | probably easy-as-pie |
| 18:15.02 | brlcad | care to make a patch? :) |
| 18:15.05 | yukonbob | it makes sense to have it do that, as it's useful, and inline w/ the behaviour of others |
| 18:15.20 | yukonbob | brlcad: I'm looking for relevant code right now ;) |
| 18:57.35 | *** join/#brlcad Z80-Boy (i=clock@77-56-77-7.dclient.hispeed.ch) |
| 18:57.58 | Z80-Boy | brlcad: I maked after the ./configure you recommended for profiling and got this: |
| 18:58.48 | Z80-Boy | /bin/sh ../../libtool --silent [...] -o btclsh [...] -lm ../../src/libtermio/libtermio.la -lm |
| 18:58.56 | Z80-Boy | /usr/X11R6/lib/libXft.so.6.0: undefined reference to `putchar' |
| 18:58.56 | Z80-Boy | /usr/X11R6/lib/libXft.so.6.0: undefined reference to `rand' |
| 19:00.25 | Z80-Boy | make src/rt/rt actually says nothing to make so I tried profiling. I ran rt ronja.g head and then |
| 19:00.28 | Z80-Boy | clock@kestrel:~/brlcad/src/rt$ gprof ./rt |
| 19:00.28 | Z80-Boy | gprof: ./rt: bad format |
| 19:00.39 | brlcad | you sure do guess a lot ;) |
| 19:00.49 | brlcad | (when you run into problems in particular) |
| 19:01.07 | brlcad | src/rt/rt is not a binary |
| 19:01.14 | Z80-Boy | oh :) |
| 19:01.30 | brlcad | it needs to be installed, or you'd have to build static |
| 19:01.49 | Z80-Boy | so now make install? |
| 19:01.49 | brlcad | those symbols from Xft are bizarre, those are like -lc symbols |
| 19:01.58 | brlcad | did the libraries build? |
| 19:02.02 | brlcad | and did you rebuild rt |
| 19:02.05 | Z80-Boy | which libraries? |
| 19:02.08 | brlcad | all of them |
| 19:02.17 | Z80-Boy | all brlcad libraries? |
| 19:02.22 | brlcad | is btclsh the first failure? |
| 19:02.29 | Z80-Boy | umm I don't know |
| 19:02.35 | Z80-Boy | I restarted the build in the meantime |
| 19:02.44 | brlcad | you're making this more complicated |
| 19:02.46 | Z80-Boy | how do I figure out if the libraries built |
| 19:02.54 | Z80-Boy | I did that on the bus |
| 19:03.00 | brlcad | do a make clean |
| 19:03.02 | brlcad | then make |
| 19:03.07 | brlcad | tell me how far it gets |
| 19:03.08 | Z80-Boy | OK |
| 19:03.45 | Z80-Boy | I don't get any compile error when I do just plain ./configure |
| 19:03.56 | Z80-Boy | and btw I found another bug in raytracing |
| 19:04.28 | brlcad | no library flags change with --enable-profile |
| 19:04.55 | brlcad | so I'd bet you had an unclean build of some sort, or something else is going on |
| 19:05.45 | brlcad | "just recompiling" certainly doesn't cause that sort of error by itself |
| 19:06.01 | *** join/#brlcad minute (n=MinuteEl@silentflame/member/pdpc.base.minuteelectron) |
| 19:06.27 | brlcad | you could add -lc to the LIBS but why you need to is really odd (but perhaps necessary on your OS) |
| 19:06.51 | brlcad | try ./configure --disable-optimized --enable-profiling LIBS=-lc |
| 19:07.10 | brlcad | i saw it |
| 19:07.16 | Z80-Boy | you get e-mails? |
| 19:07.22 | brlcad | on every change |
| 19:07.25 | Z80-Boy | wow |
| 19:07.50 | Z80-Boy | -lc? like libc? |
| 19:08.09 | Z80-Boy | Why should be this specified? C program always compiles with libc doesn't? |
| 19:08.19 | Z80-Boy | unless |
| 19:08.38 | Z80-Boy | "My name is Theo de Raadt and Chuck Norris always asks me if he's allowed to turn his computer on." |
| 19:08.58 | brlcad | those are symbols that are in libc (putchar and rand) |
| 19:09.11 | Z80-Boy | yes I know |
| 19:09.21 | brlcad | yes, most OS automatically link against libc, but not all OS's .. and apparently not yours in that instance for some reason |
| 19:09.53 | brlcad | could be a libXft problem or a libtool problem or just a quirky behavior of your OS |
| 19:10.07 | brlcad | my money is on the latter |
| 19:10.15 | Z80-Boy | if a program runs on OpenBSD it runs on anything ;-) |
| 19:10.21 | Z80-Boy | Especially regarding segfaults ;-) |
| 19:20.34 | Z80-Boy | Has anyone worked with vacuum tubes? |
| 19:20.53 | Z80-Boy | Someone suggested to put a valve into the Ronja frontend to attempt reducing the noise |
| 19:21.43 | Z80-Boy | I found a valve which has comparable parameters to the FET cascodes (tetrodes) employed but then someone on the mailing list said they actually significantly improved equipment with tubes by replacing tubes with transistors |
| 19:21.56 | Z80-Boy | and he thinks the noise they observed is caused by the hot electron cloud. |
| 19:32.49 | *** join/#brlcad ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 19:32.49 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || BRL-CAD is on scan.coverity.com but not completely set up yet, more news TBD || Release 7.10.4 imminent |
| 19:51.48 | Z80-Boy | brlcad: now the compilation wiped out at the same place |
| 19:51.53 | Z80-Boy | /usr/X11R6/lib/libXft.so.6.0: undefined reference to `putchar' |
| 19:51.53 | Z80-Boy | /usr/X11R6/lib/libXft.so.6.0: undefined reference to `rand' |
| 19:52.32 | Z80-Boy | if I add -lc after -lm, it doesn't help |
| 19:53.00 | Z80-Boy | clock@kestrel:~/brlcad/src/bwish$ nm /usr/X11R6/lib/libXft.so.6.0 | grep putchar U putchar |
| 19:54.12 | Z80-Boy | if I remove --silent, it prints the gcc call it does |
| 19:54.18 | Z80-Boy | and -lc is stripped away from there again |
| 19:54.43 | Z80-Boy | if I add -lc manually there, it works. |
| 19:55.09 | Z80-Boy | what is the purpose of libtool? Break the compilation? |
| 19:55.55 | brlcad | it apparently determined that it's already added or not needed |
| 19:56.05 | brlcad | so perhaps a libtool bug + quirky system |
| 19:56.13 | brlcad | the fact that you need it at all seems stupid |
| 19:56.24 | Z80-Boy | what does it do apart from screwing up the compilation? |
| 19:57.24 | Z80-Boy | you should into the BRL-CAD README: Requirements: non-quirky system and non-buggy libtool |
| 19:57.36 | Z80-Boy | and a URL where is described how to meet these requirements for any system |
| 19:58.07 | brlcad | libtool significantly helps 95% of the time |
| 19:58.12 | brlcad | it's that 5% that's a bitch, though |
| 19:58.27 | Z80-Boy | what does it do? |
| 19:58.28 | brlcad | still worth it overall, but can be a pita when systems aren't popular |
| 19:59.20 | brlcad | it does the determination for how to successfully compile for your platform, how to make binaries and libraries correctly, how to link, where to find and put libraries, what kind of libraries, what compilation options, linker options, on and on |
| 20:00.01 | Z80-Boy | s/correctly/without_any_guarantee_of_correctness/ |
| 20:00.02 | brlcad | the sort of details like that you need -dynamiclib on some versions of mac and -dynamic on others and -rdynamic on some versions of linux, but not others, and so on |
| 20:00.38 | brlcad | it's no more error prone than any other piece of software, all the autotools have their share of annoying bugs .. if this is even a bug |
| 20:01.01 | Z80-Boy | it's a bit labour intensive to rerun libtool and then paste the gcc etc... for every program where it fails |
| 20:01.33 | brlcad | you can set it so it always blathers the gcc line |
| 20:01.59 | Z80-Boy | yeah that's what I am doing |
| 20:02.04 | brlcad | --enable-progress |
| 20:02.14 | Z80-Boy | where? |
| 20:02.18 | brlcad | configure option |
| 20:02.27 | Z80-Boy | what does it do? |
| 20:02.30 | brlcad | --enable/disable is *always* configure |
| 20:02.45 | brlcad | it enables the gcc line |
| 20:03.18 | brlcad | it's listed under ./configure --help and in detail in the INSTALL file that covers compilation |
| 20:04.22 | Z80-Boy | I should probably replace /usr/bin/gcc with a script... |
| 20:05.17 | Z80-Boy | and here we gooooo... |
| 20:05.47 | Z80-Boy | libtoll failed to implement a protection against *THAT* :) |
| 20:06.08 | Z80-Boy | and now compile the rest of brlcad with "vanilla" gcc... |
| 20:06.43 | Z80-Boy | libtool behaves like some kind of DRM which knows better what's good for you, but knows it wrong ;-) |
| 20:08.44 | Z80-Boy | They should pass a bill that replacing /usr/bin/gcc with a script is a crime. |
| 20:10.01 | brlcad | eh, trying to replace gcc with something still called gcc but doesn't behave like gcc would be "wrong" imo, I'd want it to fail in that situation :P |
| 20:10.30 | Z80-Boy | now it fails the same way in src/mged |
| 20:10.39 | brlcad | that's about as annoying as the debian folks replacing automake and autoconf with scripts (that are outright busted in various situations) |
| 20:10.43 | Z80-Boy | so we put the script back again and compile the whole brl-cad this way |
| 20:10.59 | Z80-Boy | OpenBSD has them also replaced with script |
| 20:11.00 | brlcad | how can it "now fail" in a different place? if the build stops, it's stopped |
| 20:11.32 | brlcad | if you hack at it to try and get it to continue, and don't do so correctly, it's gonna keep failing... |
| 20:11.43 | Z80-Boy | because I used the "script method" only for bwish. |
| 20:11.56 | brlcad | you really shouldn't expect any different, the problem's not been fixed |
| 20:12.26 | Z80-Boy | libtools implements AI |
| 20:12.33 | Z80-Boy | Artificial Imcompetence |
| 20:12.43 | brlcad | the bigger question is what's different between configure with no options and configure with profile options -- I still don't believe that to be the case |
| 20:13.18 | Z80-Boy | you told me to use 3 options, --profile was one of them |
| 20:13.21 | Z80-Boy | I used it like you said. |
| 20:13.27 | brlcad | --profile wasn't one of them |
| 20:14.15 | Z80-Boy | oh sorry it was --enable-profiling: |
| 20:14.19 | Z80-Boy | "21:06 <@brlcad> try ./configure --disable-optimized --enable-profiling LIBS=-lc" |
| 20:14.20 | brlcad | and that STILL doesn't answer what's *different* |
| 20:14.29 | Z80-Boy | Why did you say LIBS=-lc? |
| 20:15.12 | brlcad | somehow I bet if you ran make clean && ./configure && make that it would also fail, if that really is the only difference |
| 20:15.17 | Z80-Boy | the configure commandline options are different |
| 20:15.29 | Z80-Boy | ./configure works, ./configure --disable-optimized --enable-profiling LIBS=-lc doesn't |
| 20:15.37 | Z80-Boy | Now it compiled with the help of my "script method". |
| 20:15.53 | brlcad | have you ran a clean configure since running clean? |
| 20:15.59 | Z80-Boy | no |
| 20:16.05 | brlcad | then your statement is bogus |
| 20:16.26 | Z80-Boy | now it compiled all brlcad successfully |
| 20:16.33 | Z80-Boy | should I do make install now? |
| 20:17.06 | Z80-Boy | maybe my statement is bogus |
| 20:17.14 | Z80-Boy | and I am also said to have bad attitude, beware! |
| 20:18.20 | brlcad | you're just way too quick to jump to conclusions that have very little if any basis, with an incomplete understanding of what's going on |
| 20:18.43 | brlcad | this problem is going to keep biting if it's not figured out |
| 20:19.26 | Z80-Boy | you mean the "undefined putchar" problem on OpenBSD when profiling is turned on? |
| 20:19.45 | Z80-Boy | now I am installing... |
| 20:19.57 | brlcad | if you don't want to figure it out, that's fine by me -- I couldn't care less, it's your system -- but then you've got very little basis to be bitching about it not working right when the problem isn't fixed and you hack around it in various ways.. |
| 20:20.15 | Z80-Boy | but can I help you with some information to figure it out? |
| 20:20.16 | brlcad | s/fixed/even determined really/ |
| 20:20.20 | Z80-Boy | I don't know what I should investigate |
| 20:21.49 | brlcad | run make clean and ./configure with no options followed by make -- now that you've cleaned beforehand, see if that works -- see if -lc is on the compile line |
| 20:22.40 | Z80-Boy | OK... |
| 20:23.00 | Z80-Boy | it'll just take a while wait |
| 20:24.11 | brlcad | thanks |
| 20:24.20 | Z80-Boy | make clean done, configure running |
| 20:24.40 | brlcad | and I'll gladly hop on again too if it's non-obvious, might see something missed |
| 20:25.22 | Z80-Boy | OH, I have profiling output! |
| 20:25.27 | Z80-Boy | make clean finished, running make... |
| 20:27.00 | Z80-Boy | is the profiling affected seriously by compilation in the background? |
| 20:30.06 | brlcad | i'd wait for the compile to finish regardless |
| 20:30.26 | Z80-Boy | sure |
| 20:30.27 | brlcad | iirc, it keeps track of both cputime and wallclock, so it'd mess with the profile a little |
| 20:35.28 | Z80-Boy | both slow and fast spend most time in _mcount |
| 20:35.33 | Z80-Boy | fast spends 22% slow 27% |
| 20:36.25 | brlcad | bah |
| 20:36.43 | brlcad | then gnu folks haven't done a good job with gprof on openbsd yet |
| 20:38.23 | Z80-Boy | can you be more specific? |
| 20:38.37 | brlcad | try adding -e mcount to gprof |
| 20:38.40 | Z80-Boy | and when I ran the rt for the profiling, it sometimes crashed and sometimes not |
| 20:38.48 | Z80-Boy | produced some "*.log" files but no core files |
| 20:38.49 | brlcad | er, -E mcount |
| 20:38.55 | Z80-Boy | it begins BRL-CAD Release 7.10.3 Crash Report |
| 20:39.07 | CIA-4 | libirc: 03JeffM2501 * r271 10/trunk/libirc/include/libIRC.h: start some doxygen type comments to get the ball rolling. |
| 20:39.09 | Z80-Boy | call stack backtrace is full of question marks |
| 20:39.46 | brlcad | there should be an rt.*crash.log file |
| 20:39.56 | brlcad | can you post it somewhere? |
| 20:40.15 | Z80-Boy | unknown-13359-bomb.log unknown-1506-bomb.log |
| 20:40.30 | brlcad | okay, those |
| 20:40.37 | Z80-Boy | on the captu? |
| 20:40.46 | CIA-4 | libirc: 03JeffM2501 * r272 10/trunk/libirc/include/libIRC.h: missed the shabang! |
| 20:40.59 | brlcad | captu? |
| 20:41.06 | Z80-Boy | where is the post-large-printouts website? |
| 20:41.18 | brlcad | ~bzpastebin |
| 20:41.18 | ibot | i guess bzpastebin is http://pastebin.bzflag.bz a place to put large chunks of text to not flood a channel |
| 20:45.18 | Z80-Boy | http://pastebin.bzflag.bz/m4c2f613e |
| 20:46.50 | Z80-Boy | here is a piece of the profile for the "fast" model: http://pastebin.bzflag.bz/m59137327 |
| 20:47.48 | Z80-Boy | And here for the "slow" one: http://pastebin.bzflag.bz/m3ffb84d6 |
| 20:48.02 | brlcad | hrmph |
| 20:48.26 | brlcad | well good to see that the automatic crash reports are working, but there were absolutely no symbols in there so it's practically useless |
| 20:49.00 | brlcad | well, not absolutely -- it's a null pointer dereference of some sort that just totally blew out the stack |
| 20:49.29 | brlcad | which .. openbsd is supposed to prevent from happening too.... odd |
| 20:50.35 | Z80-Boy | I don't understand it into such details |
| 20:50.51 | brlcad | did you add the -E mcount? |
| 20:50.53 | Z80-Boy | I just know Theo's Puffy keeps burglars out of my P.C. by some stack randomization and whatever other tricks. |
| 20:50.59 | Z80-Boy | no I didn't |
| 20:51.06 | Z80-Boy | that was without -E mcount |
| 20:51.39 | brlcad | try both with "-E mcount -E _mcount" just to be sure |
| 20:51.44 | brlcad | make sure there's no mcount in the result |
| 20:51.52 | brlcad | s/result/report/ |
| 20:52.16 | Z80-Boy | now worker [1] is the number one |
| 20:54.36 | Z80-Boy | now in the bad one the calloc is the #1 |
| 20:54.39 | Z80-Boy | should I post it? |
| 20:57.19 | Z80-Boy | hey look crash again |
| 20:57.30 | Z80-Boy | Additional mem=0., #malloc=1575, #free=1318, #realloc=4 (257 retained) |
| 20:57.30 | Z80-Boy | 804dc0bu_log: write error |
| 20:57.30 | Z80-Boy | Saving stack trace to unknown-13652-bomb.log |
| 20:57.55 | Z80-Boy | no core again |
| 20:58.12 | Z80-Boy | and again stack all question marks |
| 20:58.59 | Z80-Boy | 2 more crashes in a row and 6 successful raytraces |
| 20:59.11 | Z80-Boy | Maybe it's the thread ordering? Or page ordering in memory? |
| 21:00.15 | Z80-Boy | Interesting. If I run 'find /', then the rt gets a fit of couple crashes and then subsequent runs raytrace happily |
| 21:00.22 | Z80-Boy | until I run the find / again ;-) |
| 21:01.53 | Z80-Boy | the slow model needs 15.5 sec the fast one 0.074 sec |
| 21:02.13 | Z80-Boy | That's 209x faster |
| 21:02.18 | Z80-Boy | Is it normal? Just a cutaway... |
| 21:03.18 | *** join/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) |
| 21:03.33 | Z80-Boy | By CPU time the difference is 214x |
| 21:07.20 | Z80-Boy | OK make finished OK |
| 21:07.39 | Z80-Boy | There are lots of gcc commandlines in the compile, one example: |
| 21:08.33 | Z80-Boy | I can't find one it's all libtool |
| 21:15.24 | CIA-4 | BRL-CAD: 03lbutler * 10brlcad/src/conv/intaval-g.py: a script that parses INTAVAL files into Tcl that mged can read. |
| 21:18.12 | *** join/#brlcad ibot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 21:18.12 | *** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || http://sf.net/projects/brlcad || BRL-CAD is on scan.coverity.com but not completely set up yet, more news TBD || Release 7.10.4 imminent |
| 21:19.53 | ``Erik | um, the whole bsd family tends to be finicky about profiling shtuff |
| 21:20.16 | ``Erik | all the symbols are provided in libraries (libc_r.so libc_p.so etc) |
| 21:20.21 | ``Erik | but gcc is... gooby |
| 21:22.14 | CIA-4 | libirc: 03brlcad * r273 10/trunk/libirc/include/ (6 files): some more doxygenification |
| 21:23.34 | brlcad | Z80-Boy: no, I think you pinpointed a really bad inefficiency (calloc) that shouldn't be happening .. that's the glory of profiling :) |
| 21:23.46 | brlcad | just need to see the profile myself to see where it's coming from |
| 21:24.26 | brlcad | i'm about to run out for a few hours, so I'll have to catch up |
| 21:25.05 | ``Erik | is the 'cutaway' using a "half" primitive? I'm under the impression that 'half' is horrible for performance |
| 21:25.34 | ``Erik | (also, I've never thought of obsd as an efficient for fast os... it's the "safe" one...) |
| 21:32.14 | Z80-Boy | what's the story morning glory |
| 21:39.13 | Z80-Boy | bool.c: " * The list is terminated with a NULL pointer, |
| 21:39.13 | Z80-Boy | <PROTECTED> |
| 21:39.14 | Z80-Boy | " |
| 21:39.19 | Z80-Boy | Sounds like could be optimized... |
| 21:42.51 | Z80-Boy | shoot.c also contains one calloc which smells like could be optimized |
| 21:43.55 | Z80-Boy | basically any calloc is suspicious |
| 21:44.04 | Z80-Boy | what's the point of having a memory block zeroed out? |
| 21:44.15 | Z80-Boy | If you want to write into part of it then do malloc, write, and zero out the rest |
| 21:45.05 | Z80-Boy | If you don't write then you can replace any read into the block with 0 |
| 21:45.17 | Z80-Boy | so you don't need this block to be actually present anywhere and can just forget it |
| 21:50.14 | *** join/#brlcad iraytrace (n=iraytrac@cocoa.sci.utah.edu) |
| 22:22.23 | *** join/#brlcad PrezKennedy (i=Matt@74.86.45.130) [NETSPLIT VICTIM] |
| 22:22.23 | *** join/#brlcad yukonbob (n=yukonbob@198.235.198.234) [NETSPLIT VICTIM] |
| 22:23.02 | *** join/#brlcad tarzeau (i=gurkan@bee.ethz.ch) [NETSPLIT VICTIM] |
| 22:23.02 | *** join/#brlcad Maloeran (n=maloeran@glvortex.net) [NETSPLIT VICTIM] |
| 22:23.07 | *** join/#brlcad SWPadnos (n=Me@emc/developer/SWPadnos) [NETSPLIT VICTIM] |
| 22:23.59 | *** join/#brlcad poolio_ (n=poolio@c-71-206-215-46.hsd1.pa.comcast.net) [NETSPLIT VICTIM] |
| 22:23.59 | *** join/#brlcad dtidrow_work (n=dtidrow@host169.objectsciences.com) [NETSPLIT VICTIM] |
| 22:24.02 | *** join/#brlcad ChanServ (ChanServ@services.) |
| 22:24.02 | *** mode/#brlcad [+o ChanServ] by irc.freenode.net |
| 22:24.03 | *** join/#brlcad AchiestDragon (n=david@80.177.175.151) |
| 22:24.49 | *** join/#brlcad starseeker (n=CY@ip72-218-16-62.hr.hr.cox.net) [NETSPLIT VICTIM] |
| 22:24.49 | *** join/#brlcad b0ef (n=b0ef@062016141081.customer.alfanett.no) [NETSPLIT VICTIM] |
| 22:25.17 | *** join/#brlcad brlcad (n=sean@pdpc/supporter/silver/brlcad) |
| 22:25.17 | *** join/#brlcad thing0 (n=ric@203-59-138-206.perm.iinet.net.au) [NETSPLIT VICTIM] |
| 22:25.17 | *** join/#brlcad MinuteElectron (n=MinuteEl@silentflame/member/pdpc.base.minuteelectron) [NETSPLIT VICTIM] |
| 22:25.17 | *** join/#brlcad Z80-Boy (i=clock@77-56-77-7.dclient.hispeed.ch) [NETSPLIT VICTIM] |
| 22:25.17 | *** join/#brlcad CIA-4 (i=cia@208.69.182.149) [NETSPLIT VICTIM] |
| 22:25.17 | *** join/#brlcad ``Erik (i=erik@c-69-250-155-85.hsd1.md.comcast.net) [NETSPLIT VICTIM] |
| 22:25.17 | *** mode/#brlcad [+o brlcad] by irc.freenode.net |
| 22:56.48 | ``Erik | heh |