IRC log for #brlcad on 20120125

16:20.39*** join/#brlcad ibot (~ibot@rikers.org)
16:20.39*** topic/#brlcad is BRL-CAD Open Source Solid Modeling || http://brlcad.org || logs: http://ibot.rikers.org/%23brlcad/ || BRL-CAD release 7.22.0 is forthcoming || fixing all our Coverity Scan Initiative defects
16:21.23``Erikyour compiler isn't liking the __extension__ bit, which version of gcc is that?
16:21.59jordisayolgcc version 4.5.1
16:22.19``Erikand your libc headers are the most recent?
16:22.45jordisayolI think so
16:22.58jordisayolI've system up to date
16:25.24jordisayolglibc 2.11.3-12.21.1
16:27.29starseeker``Erik: awesome
16:27.46starseekerhope they produce it
16:28.47``Eriklacks much in the way of linux machines, is going to try compiling with a new gcc on fbsd to see if its' a change in gcc O.o
16:29.50``Erikbut this is looking like an opensuse issue to me, your gcc doesn't like your system headers :/ any program that uses FD_SET() would be throwing this issue (might be some flags we use triggering it)
16:32.10jordisayolok. I've compiled all previous rpm packages on this system so, is new the use of this function on brlcad?
16:33.11jordisayolif not, then is a new bug on opensuse glibc, isn't it?
16:33.49``Erikhm, it's an old use of that macro in BRL-CAD, only a couple changes in that file in the last few months
16:34.11``Erikone on dec 15, the other on nov 29
16:34.29``Erikbuilt fine using gcc 4.7 on fbsd 8.2, ia32
16:35.05jordisayolno no, it build fine on same opensuse 64-bit!!!
16:35.49``Erikyeah, it's an assembly macro, so it's platform specific...
16:36.13``Erikit's coming from /usr/include/asm-*/posix_types.h
16:36.35``Erikso on 64, you get asm-x86_64, on 32 you get asm-i386... different files
16:37.09jordisayolcan i do something to override it?
16:37.29``Eriknot to my knowledge, but I'm not a linux guy and have never logged into an opensuse box
16:37.52jordisayolor better, when will be released the next brlcad version?
16:39.03``Eriksorry, dude, I'm out of ideas, maybe the opensuse guys can help? the __extension__ on line 53 of your paste is the bit causing the issue, and that's from /usr/include/asm-i386/posix_types.h *shrug*
16:39.56``Eriknext BRL-CAD release should be fairly soon, but I don't think it'll fix stuff or ya :/
16:40.10``Eriks/ or/ for/
16:41.03jordisayolI ask you just to know if there is enough time to opensuse crew correct this
16:41.19starseekerhow much time would they need?
16:41.29jordisayolno idea
16:41.35starseekerwe won't be putting a release out in the next few days - there's too much to do
16:42.07jordisayolI'm not an opensuse user. I just use it in a virtualbox machine to build brlcad rpm packages
16:42.39starseekermy advice would be to contact them and see what they say
16:43.06*** join/#brlcad packrat (~packrator@c-98-209-146-133.hsd1.mi.comcast.net)
16:43.56jordisayolok, i'll try, many thanks!
16:44.22``Erikthey may need to see your compile line, it might be something funky like -msse3 or something causing it to trigger
17:08.32CIA-48BRL-CAD: 03brlcad * r49068 10/brlcad/trunk/src/ (13 files in 3 dirs): (log message trimmed)
17:08.32CIA-48BRL-CAD: Can't take it any longer. Doggy bag all the gcc vomit about SCL's qualified type errors. Reduces the build log by about 16MB, 78k warning lines, approx 765 unique instances.
17:08.32CIA-48BRL-CAD: Quell them all with a teeny little one-liner bit of scripting based on the build log output:
17:08.32CIA-48BRL-CAD: for match in `grep qualifiers build.log | sort | uniq | cut -d: -f1,2` ; do
17:08.33CIA-48BRL-CAD: export file="`echo $match | cut -d: -f1`" ; export line="`echo $match | cut -d:
17:08.33CIA-48BRL-CAD: -f2`" ; export pline="`expr $line - 1`" ; if test "x`sed -n ${pline}p
17:08.34CIA-48BRL-CAD: $file |grep const`" = "x" ; then sed -n ${line}p $file ; sed
17:08.36*** join/#brlcad merzo (~merzo@248-93-200-46.pool.ukrtel.net)
17:10.33CIA-48BRL-CAD: 03brlcad * r49069 10/brlcad/trunk/src/other/step/src/clstepcore/ExpDict.h:
17:10.33CIA-48BRL-CAD: more warning quellage. shouldn't pass string constants to char* parameters as
17:10.33CIA-48BRL-CAD: they should be considered immutable/const. make SelectTypeDescriptor()
17:10.33CIA-48BRL-CAD: constructor take a const char* instead, fortunately TypeDescriptor's constructor
17:10.33CIA-48BRL-CAD: is correct.
17:12.12CIA-48BRL-CAD: 03bob1961 * r49070 10/brlcad/trunk/src/mged/polyif.c: Variable declarations need to be at the top of the code block.
17:22.41CIA-48BRL-CAD: 03brlcad * r49071 10/brlcad/trunk/src/other/step/src/clstepcore/ (8 files):
17:22.41CIA-48BRL-CAD: might make merging a bit more difficult or (more likely) repeats work already
17:22.41CIA-48BRL-CAD: done in the github tree, but helps ease our sanity in the meantime. eliminate a
17:22.41CIA-48BRL-CAD: slew of gcc warnings about string literals being passed as char* parameters when
17:22.41CIA-48BRL-CAD: they should be treated immutable. mark them all const.
17:23.02brlcadthat takes care of the majority of the step vomit
17:23.20brlcadthe remaining warnings look like they're other valid step-g issues
17:24.58brlcadnow the warning log has but 520 or so lines in it (instead of 78k lines)
17:26.06brlcadjordisayol: what does man FD_SET list for headers?
17:27.21jordisayolbrlcad: sorry, I don't understand you
17:27.45brlcadif you run "man FD_SET" it should give a usage man page
17:27.59brlcadit should list one or more #include headers near the top
17:28.00brlcadlike
17:28.01brlcad#include <sys/select.h>
17:28.03jordisayolnop, no man pacge for it
17:28.11brlcadhm, okay, what about man select
17:28.38jordisayolnot too
17:29.01brlcadno man page for select?
17:29.25brlcadsounds like you're missing manual pages
17:29.40jordisayoljust a moment, i just use opensuse to build brlcad
17:29.51jordisayolmaybe it's not installed
17:30.17brlcadno matter, found it online http://manpages.unixforum.co.uk/man-pages/linux/opensuse-10.2/0p/select.h-man-page.html
17:31.29jordisayolnow i got man FD_SET
17:31.56brlcadgrep HAVE_SYS_SELECT_H include/*
17:32.03brlcadfrom your build directory
17:32.56jordisayolinclude/brlcad_config.h:#define HAVE_SYS_SELECT_H 1
17:32.56jordisayolinclude/brlcad_config.h.in:#cmakedefine HAVE_SYS_SELECT_H 1
17:32.56jordisayolinclude/bselect.h:#ifdef HAVE_SYS_SELECT_H
17:33.07brlcadhm, okay that's good
17:34.41brlcadjordisayol: try this .. edit include/bselect.h and add this line before the #ifdef HAVE_SYS_SELECT_H line
17:35.06brlcad#include <sys/types.h>
17:35.13brlcad#include <signal.h>
17:35.17brlcad#include <time.h>
17:35.26brlcadthose three lines
17:35.40brlcadsee if that makes any difference at all
17:35.51jordisayolok
17:38.17*** join/#brlcad Stattrav (~Stattrav@61.12.114.82)
17:38.17*** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav)
17:38.37jordisayolbrlcad: it appears that pass the error point
17:38.58jordisayolstill compiling...
17:39.56jordisayolit will take a wile (compiling in a virtualbox machine with just one cpu)
17:40.13brlcadwow
17:40.26brlcadyeah, that's a busted system header then
17:40.43brlcadremove the signal.h and time.h lines and retry
17:40.56jordisayolok
17:41.24jordisayolohhh, sorry, the error still there :-/
17:41.25brlcadthen assuming that works, remove the sys/types.h line just to verify that's what makes it work
17:41.31brlcadhm, okay
17:46.03brlcad``Erik: nifty -E -o trick, handy thx
17:50.18brlcadjordisayol: you could try a newer version of gcc, one of the 4.6 line
17:51.51jordisayolI'll try, I'm not comfortable on opensuse...
17:55.30*** join/#brlcad juan_man (~quassel@186.136.169.88)
17:55.58*** join/#brlcad juan_man (~quassel@unaffiliated/juanman)
17:56.52jordisayolbrlcad: anyway, on opensuse 11.4 64-bit with gcc 4.5.1 , properly compiles
17:57.20CIA-48BRL-CAD: 03starseeker * r49072 10/brlcad/trunk/src/other/libpng/CMakeLists.txt: If we're conditionalizing the non-WIN32 files on generation, need to also conditionalize the related install rules.
18:00.17CIA-48BRL-CAD: 03starseeker * r49073 10/brlcad/trunk/src/other/libpng/CMakeLists.txt: ws
18:20.29CIA-48BRL-CAD: 03n_reed * r49074 10/brlcad/trunk/src/libbu/test_sscanf.c: Using macros to condense numeric conversion tests, now in their own routine. Moving on to more exotic conversions.
18:22.39jordisayolbrlcad: I've upgraded to gcc 4.6.2 (from opensuse 12.1) and it appears to work
18:22.46jordisayolstill compiling
18:32.39jordisayolok, successfully compiled
18:43.32CIA-48BRL-CAD: 03starseeker * r49075 10/brlcad/trunk/src/nirt/command.c: Send the default coordinate printouts to stdout - really should be handled like 'normal' nirt reporting...
18:44.54brlcadjordisayol: awesome
18:45.17brlcaddid you see the forum message about the opensuse rpm not working?
18:45.45jordisayolnop :-(
18:45.57brlcadhttps://sourceforge.net/projects/brlcad/forums/forum/362510/topic/4972540
18:46.11brlcadthought that might be why you were looking into opensuse :)
18:49.09CIA-48BRL-CAD: 03starseeker * r49076 10/brlcad/trunk/src/nirt/command.c: another stdout fprintf change...
18:49.17jordisayolno no :-) just test to create deb/rpm packages randomly
18:57.59brlcadany idea what that error might be?
18:58.25jordisayolnot really
18:59.05brlcadI know ogl_interface is a symbol in our libfb library, only compiled if you enable opengl
18:59.27brlcadis the rpm he's using supposed to have opengl on or off?
18:59.46jordisayolon
19:00.15brlcadso then the fact that there is a reference to ogl_interface somewhere is correct
19:00.25brlcadwhy it can't find it becomes the issue
19:00.47brlcadmaybe he has another library on his system named libfb
19:00.49jordisayolso probably is a problem with his opengl. do you remember that some other person got problems on opensuse too?
19:01.27brlcadnope
19:02.17jordisayolwell, this guy got problems running mged/archer
19:02.46jordisayoland finally he wasn't able to run  glxgears on his system
19:03.05brlcadright, so that one sounds like an opengl issue
19:03.16brlcadnot finding ogl_interface isn't, though
19:04.28jordisayolmissing dependencies on rpm?
19:05.02brlcadyou don't specify object files in the rpm, so not likely
19:05.19brlcadthat's just a specific symbol inside libfb that is referenced by libdm and mged
19:06.47brlcadwhat I'm betting is happening is that it's loading mged, begins loading libs, loads libfb but finds *another* system libfb instead of ours, loads libdm, see's the ogl_interface symbol, gets through loading all the of rest of the libs and has no way to resolve it
19:08.05jordisayolaha
19:08.39brlcadmaybe even another compile of BRL-CAD
19:09.20jordisayolaha
19:11.54*** join/#brlcad Stattrav (~Stattrav@61.12.114.82)
19:11.54*** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav)
19:17.12jordisayolbrlcad: if there is a /usr/lib/libdm.so and a /usr/brlcad/lib/libdm.so, which one is used by mged?
19:21.41``Erik'ldd' will tell you. "ldd /path/to/mged"
19:27.31jordisayolI've installed libdm0 (installs /liblibdm.so.0.0.4) and ldd /usr/brlcad/mged returns:
19:27.31jordisayol...
19:27.31jordisayollibdm.so.19 => /usr/brlcad/lib/libdm.so.19 (0x00007f38e9ac9000)
19:27.31jordisayol...
19:29.45jordisayols:/liblibdm.so.0.0.4:/lib/libdm.so.0.0.4:
19:35.55brlcadbinaries are usually compiled with linker information pointing to where the library is expected to be
19:36.44brlcadso if you don't install where it was compiled, LD_LIBRARY_PATH becomes necessary or modifications to ld.config
19:37.08brlcadif you do install where it was compiled for, you don't generally need anything
19:38.24jordisayolrpm package always installs libs at /usr/brlcad/lib
19:39.02brlcadso then it's back to him having a libfb just somewhere else on his system and LD_LIBRARY_PATH is pointing to it
19:42.49jordisayolI've to go
19:42.53``Erikรด/cl
19:45.34brlcadcya jordisayol
19:45.37brlcadthanks!
19:47.25*** join/#brlcad pawleeq (~pawleeq@core1.humlnet.cz)
19:47.30pawleeqhello
19:47.55pawleeqhere is another czech article about brlcad: http://www.abclinuxu.cz/clanky/brl-cad-pokrocile-techniky-modelovani
19:52.45brlcadpawleeq: that is awesome!
19:52.58brlcadand you even figured out how to work that terrible pattern tool gui :D
19:53.24pawleeqyeah and it made me to think about few improvements ...
19:53.33pawleeqi will write a hot report soon a let you know
19:53.41brlcadthe "clone" command is a lot better, at least easier to use
19:53.42pawleeq* a > and
19:53.51brlcadthough it doesn't do the spherical/cylindrical patterns
19:53.58brlcadcool
19:54.19brlcadit'd be great to have those tutorials in docbook format :)
19:54.38pawleeqone of the readers asked specially about cylindrical patterns so I did it
19:54.40brlcadthen we could get some english translations going and get them up into our docs
19:55.36pawleeqI can translate it, i even think about real book, but these articles are fortunately keeping my feet on the ground :)
20:18.33brlcadpawleeq: do you know docbook/xml?  or html?
20:19.36pawleeqI know html
20:20.49brlcaddocbook/xml is very very similar -- if you're interested, I could pull together a template example
20:21.09brlcadfrom docbook, we can autogenerate html, pdf, rtf, etc
20:21.52brlcadwhen was your first article published?
20:22.53pawleeqNovember 11th
20:23.32pawleeqok send me a template and I will fill it with english translation of those articles (or just the articles you want)
20:24.07brlcadboth english and czech would be best :)
20:24.14pawleeqok
20:24.18brlcadwe can track translations
20:24.47brlcadthat way, we might get others willing to help translate your articles to other languages as well
20:25.44pawleeqok :)
20:30.47CIA-48BRL-CAD: 03brlcad * r49077 10/brlcad/trunk/AUTHORS: credit pavel jansa (pawleeq) with his numerous brl-cad tutorial articles (in czech): http://www.abclinuxu.cz/autori/pavel-jansa
20:32.47CIA-48BRL-CAD: 03brlcad * r49078 10/brlcad/trunk/src/libbn/noise.c:
20:32.47CIA-48BRL-CAD: clean up the magic number handling to be consistent with our uint32_t magic
20:32.47CIA-48BRL-CAD: type. expand the size of the hash table from shorts to that as well and do some
20:32.47CIA-48BRL-CAD: minor refactoring so that the table size (4096) is defined in just one place and
20:32.47CIA-48BRL-CAD: everything keys off of it.
20:37.06CIA-48BRL-CAD: 03brlcad * r49079 10/brlcad/trunk/src/libbu/badmagic.c:
20:37.06CIA-48BRL-CAD: not clear what is exactly being checked but the magic number alignment checking
20:37.06CIA-48BRL-CAD: looks to be incorrect now that we're using uint32_t's everywhere. getting
20:37.06CIA-48BRL-CAD: misalignment on big-endian 64-bit system in shaders regression, fixed by
20:37.06CIA-48BRL-CAD: checking our magic type.
20:42.04*** join/#brlcad Stattrav (~Stattrav@61.12.114.82)
20:42.05*** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav)
20:52.17CIA-48BRL-CAD: 03n_reed * r49080 10/brlcad/trunk/src/libbu/sscanf.c: Forgot that we need to search for closing brace of character classes. Was causing truncated conversion specification.
21:01.04CIA-48BRL-CAD: 03starseeker * r49081 10/brlcad/trunk/src/nirt/sfiles/csv-gap.nrt: Stray commas in csv-gap formatting were causing error when specifying -f csv-gap in a nirt run.
21:48.10CIA-48BRL-CAD: 03n_reed * r49082 10/brlcad/trunk/src/libbu/sscanf.c: Doh! Accidently overwrote conversion type var before switching on it.
21:52.38CIA-48BRL-CAD: 03starseeker * r49083 10/brlcad/trunk/ (NEWS src/nirt/if.c): Attribute printing was haywire in nirt, printing ALL attributes and ignoring the list. Now printing only specified attributes.
21:56.44n_reedhah, accidentally used accidently
22:10.34CIA-48BRL-CAD: 03starseeker * r49084 10/brlcad/trunk/src/nirt/command.c: err -> stdout for nirt default units reporting.
22:14.37CIA-48BRL-CAD: 03n_reed * r49085 10/brlcad/trunk/src/libbu/test_sscanf.c: Add some string-type tests now that character-class conversion is debugged.
22:16.01CIA-48BRL-CAD: 03starseeker * r49086 10/brlcad/trunk/ (NEWS src/nirt/nirt.c src/nirt/nirt.h): Add -h option to nirt to suppress header, couple other tweaks.
22:21.31CIA-48BRL-CAD: 03starseeker * r49087 10/brlcad/trunk/ (TODO regress/CMakeLists.txt regress/nirt.sh): Add regression testing for nirt based on the nirt DocBook manual.
22:34.57CIA-48BRL-CAD: 03brlcad * r49088 10/brlcad/trunk/doc/docbook/articles/en/ (CMakeLists.txt TEMPLATE.xml): add an initial template that demonstrates some of the basic docbook/xml elements for adding new articles.
22:57.01*** join/#brlcad pawleeq (~pawleeq@core1.humlnet.cz)
23:18.25brlcadpawleeq: http://brlcad.svn.sourceforge.net/viewvc/brlcad/brlcad/trunk/doc/docbook/articles/en/TEMPLATE.xml?view=log
23:18.42brlcadthere's a template you can use, it'll generate something like this: http://brlcad.org/~sean/tmp/TEMPLATE.html
23:19.14pawleeqok thank you
23:19.34brlcadit's in svn, so you can copy that file, update the doc/docbook/articles/en/CMakeLists.txt file with your article(s)
23:19.49brlcadand when you run "make", it'll generate the html
23:20.14brlcaddon't worry about how the html looks exactly, it's more that it's structurally all there described and easy enough to read
23:20.23brlcadif that wasn'
23:20.36brlcadif that wasn't clear, folks in here can easily help ;)
23:24.40pawleeqok I will ask if I get unsure about something, and there always is google :) I am going to bed, so good bye everybody
23:55.32*** join/#brlcad starseeker (~starseeke@BZ.BZFLAG.BZ)
23:57.03*** join/#brlcad piksi (piksi@pi-xi.net)
23:58.41*** join/#brlcad ChanServ (ChanServ@services.)
23:58.41*** mode/#brlcad [+o ChanServ] by pratchett.freenode.net

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