| 00:13.10 | Notify | 03BRL-CAD:brlcad * 54861 brlcad/trunk/src/mged/mged.c: remove the debug printing to the non-portable /tmp/stdout file, makes several bogus assumptions. also be sure to remove our bu_log hook that redirects output to the GUI when we're closing the application. there may be other calls to bu_log() that will crash after Tcl is shut down. |
| 00:18.56 | Notify | 03BRL-CAD:brlcad * 54862 brlcad/trunk/src/mged/mged.h: declare gui_output |
| 00:19.16 | Notify | 03BRL-CAD:brlcad * 54863 brlcad/trunk/src/mged/cmd.c: moved gui_setup to mged.h so no longer need to declare here |
| 00:21.32 | Notify | 03BRL-CAD:brlcad * 54864 brlcad/trunk/src/mged/mged.c: looks like LOGFILE is no longer used, which is good because this looks like a total non-portable hack |
| 00:23.04 | Notify | 03BRL-CAD:brlcad * 54865 brlcad/trunk/src/mged/mged.c: slow commit, wasn't supposed to get this yet |
| 01:17.30 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 01:30.54 | *** join/#brlcad n_reed_ (~molto_cre@66-118-151-70.static.sagonet.net) | |
| 01:40.56 | Notify | 03BRL-CAD:brlcad * 54866 brlcad/trunk/src/mged/mged.c: restore stdout and stderr when we're done with the GUI by capturing them with a dup() call and then later restoring back with dup2(). this fixes a problem where an atexit() callback couldn't write out anything during shutdown (SIGPIPE) because Tcl had destroyed the channel we set up. document and descope a couple of the globals while we're at it. untested, but |
| 01:40.59 | Notify | this should work fine on windows as well because dup/dup2 are c89 and msdn-supported. begging for some _WIN32 cleanup consolidation (the duplication seems unnecessary). |
| 01:44.11 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 01:54.48 | *** join/#brlcad FLOSSrookie (~brian@107-200-34-111.lightspeed.tulsok.sbcglobal.net) | |
| 01:55.06 | FLOSSrookie | Anyone active to assist? |
| 01:56.44 | FLOSSrookie | I was kinda forced to move to another distro. I was using Fedora but it was far too unstable for even basic usage. I am now using centOS but the old instructions from last time are no longer applicable. I will have to build from source this time unless one of the fedora RPMs will work. Will a fedora RPM work for centOS? |
| 01:57.20 | FLOSSrookie | I'll re-ask tomorrow if no one is active. |
| 02:04.53 | *** part/#brlcad FLOSSrookie (~brian@107-200-34-111.lightspeed.tulsok.sbcglobal.net) | |
| 02:17.17 | Notify | 03BRL-CAD:brlcad * 54867 brlcad/trunk/src/libbu/heap.c: make the debug printing conditional on bu_debug being set to something |
| 02:21.42 | brlcad | mm, have to wait longer than that |
| 02:49.27 | Notify | 03BRL-CAD:brlcad * 54868 brlcad/trunk/src/liboptical/sh_light.c: not just rare but unlikely that there are thousands of light sources, so let them be allocated on the system heap. pretty big struct (295 bytes) |
| 03:04.44 | Notify | 03BRL-CAD:brlcad * 54869 brlcad/trunk/src/libbu/heap.c: support a wider range of allocation request sizes (1-1024) but keep page sizes at 1MB per size. add the ability to bomb on out-of-range requests so they can be traced. |
| 03:08.09 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 03:31.58 | Notify | 03BRL-CAD:brlcad * 54870 brlcad/trunk/include/conf/PATCH: significant third-party updates, bump patch to keep track |
| 03:35.44 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 03:55.17 | Notify | 03BRL-CAD:brlcad * 54871 brlcad/trunk/include/bu.h: rt_avail_cpus() was renamed to bu_avail_cpus() a long time ago, update comment along with bu_hist_range() |
| 04:44.17 | *** join/#brlcad merzo (~merzo@95.132.33.92) | |
| 08:02.18 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 08:03.55 | *** join/#brlcad luca79 (~luca@net-188-216-230-48.cust.dsl.vodafone.it) | |
| 09:06.18 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 09:14.17 | *** join/#brlcad luca79 (~luca@net-188-216-230-48.cust.dsl.vodafone.it) | |
| 09:30.03 | *** join/#brlcad jordisayol (~jordisayo@unaffiliated/jordisayol) | |
| 09:31.55 | *** join/#brlcad merzo_ (~merzo@user-94-45-58-138-1.skif.com.ua) | |
| 10:34.46 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 11:15.36 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 12:13.42 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 12:28.52 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 13:08.35 | Notify | 03BRL-CAD:starseeker * 54872 brlcad/trunk/src/other/libpng.dist: Add config.h.in to png dist file. |
| 13:20.51 | *** join/#brlcad milamber (~devlin@d118-75-244-176.try.wideopenwest.com) | |
| 13:53.46 | Notify | 03BRL-CAD:carlmoore * 54873 brlcad/trunk/src/mged/mged.c: fix spelling in a comment |
| 14:07.04 | *** join/#brlcad luca79 (~luca@net-188-216-230-48.cust.dsl.vodafone.it) | |
| 14:13.43 | Notify | 03BRL-CAD:n_reed * 54874 brlcad/trunk/src/libged/draw.c: fix leaked path memory in ged_redraw |
| 14:22.17 | Notify | 03BRL-CAD:brlcad * 54875 brlcad/trunk/src/util/pix-png.c: undo apparent accidental inclusion of pix-png.c in r54858. it made help options not work (and incorrectly spits an error about not outputting to a tty). |
| 14:39.36 | *** join/#brlcad merzo_ (~merzo@user-94-45-58-138-1.skif.com.ua) | |
| 14:51.27 | Notify | 03BRL-CAD:n_reed * 54876 (brlcad/trunk/src/other/step/src/clstepcore/STEPaggregate.h brlcad/trunk/src/other/step/src/fedex_plus/classes.c and 3 others): Facilitate merge work by removing the const_ typedef return types introducded in r49683. It's not clear that returned values need const qualification, or that they are assigned to lvalues that preserve that qualification, so these const_ types are presently an |
| 14:51.28 | Notify | elaborate way to accomplish nothing, and moreover have caused issues parsing some express inputs. Until someone can authoritatively determine what values should and should-not be considered read-only, we'll take a much simpler approach to suppressing misued qualifier warnings by just omitting the erroneous application of const to generated return types. |
| 14:59.10 | Notify | 03BRL-CAD:r_weiss * 54877 (brlcad/trunk/src/other/tk/win/wish.exe.manifest =================================================================== and 34 others): This file is needed for windows to build. |
| 15:02.01 | Notify | 03BRL-CAD:n_reed * 54878 brlcad/trunk/src/other/step/src/fedex_plus/classes.c: put attrFlags declarations in preprocessor conditionals to match wrapped uses; avoids unused variable warnings |
| 15:15.33 | Notify | 03BRL-CAD:brlcad * 54879 (brlcad/trunk/src/other/step/src/express/expparse.y brlcad/trunk/src/other/step/src/express/generated/expparse.c): looks like a stack size of 0 means grow dynamically on the heap instead of static allocation. array is going to be faster, but it's more critical to not hit a compile-time limit. by default, grow the dynamic allocation from 256 in powers of two. |
| 15:16.34 | Notify | 03BRL-CAD:brlcad * 54880 brlcad/trunk/src/other/lemon/lempar.c: set the initial dynamic allocation to 256 for better memory alignment and doubling in powers of two |
| 15:40.59 | maths22 | Is the new server today? |
| 15:45.09 | Notify | 03BRL-CAD:brlcad * 54881 (brlcad/trunk/src/other/step/cmake/Generated_Source_Utils.cmake brlcad/trunk/src/other/step/src/express/generated/verification_info.cmake): if we only strip evil carriage returns, we can use md5sum on the file directly to verify and update them. |
| 15:45.23 | brlcad | maths22: good question -- it's supposed to be |
| 15:45.30 | brlcad | I believe the scheduled time is 11pm |
| 15:56.00 | maths22 | Does that mean our running processes will be killed? |
| 15:56.04 | maths22 | i.e. irssi |
| 16:00.26 | brlcad | heh, kind of hard to pull the hard drive out and put it into another machine with new hardware while keeping the processes running ;) |
| 16:01.01 | brlcad | this isn't enterprise sun solaris hardware |
| 16:04.28 | Notify | 03BRL-CAD:brlcad * 54882 brlcad/trunk/src/other/step/src/express/generated/verification_info.cmake: missing a 9, typo apparently |
| 16:07.17 | maths22 | so the physical harddrive moves; the data is not transfered to a new hdd. |
| 16:11.56 | brlcad | right |
| 16:12.50 | brlcad | I didn't want to go through yet another drawn out filesystem migration, that was painful |
| 16:13.36 | maths22 | was that what happened last time? |
| 16:13.47 | brlcad | if they can just move the drive, everything should stay the same except for maybe a one-liner if the network interface name changes |
| 16:14.04 | maths22 | also, will the server ip change? |
| 16:14.07 | brlcad | yes, the DIY route |
| 16:14.13 | brlcad | no, it shouldn't |
| 16:14.45 | brlcad | last time IPs changed, major OS version changes, nothing was compatible |
| 16:15.20 | brlcad | this is close enough hardware and ``Erik has the operating system enough up-to-date that it should just boot up happy |
| 16:15.41 | brlcad | and if not, I'm still paying them to get it working |
| 16:15.51 | brlcad | so more expensive, but way faster than relying on my time |
| 16:17.31 | maths22 | Do they perform the network interface changes locally? |
| 16:22.12 | maths22 | Also, can the hardware support two hdds? |
| 16:26.43 | Notify | 03BRL-CAD:r_weiss * 54883 brlcad/trunk/src/proc-db/molecule.c: Fix a windows build error. |
| 16:31.33 | *** join/#brlcad hsrai (~hsrai@202.164.53.116) | |
| 16:44.03 | brlcad | maths22: the hwardware can support two hds, but will only have the one (we'll have offsite backups) |
| 16:44.14 | brlcad | the network interface change is just a line in a text file |
| 16:49.41 | ``Erik | 2 lines in rc.conf, iirc |
| 17:37.15 | Notify | 03BRL-CAD:bob1961 * 54884 brlcad/trunk/src/tclscripts/archer/ArcherCore.tcl: ----------- |
| 17:55.12 | *** part/#brlcad witness (uid10044@gateway/web/irccloud.com/x-kmepvsuuoqzwxsmb) | |
| 18:07.14 | *** join/#brlcad FLOSSrookie (~brian@107-200-34-111.lightspeed.tulsok.sbcglobal.net) | |
| 18:07.57 | FLOSSrookie | I move to another distro. I was using Fedora but it was too unstable for even basic usage. I am using centOS but the old instructions from last time I was here are no longer applicable. I will have to build from source this time unless one of the fedora RPMs will work. Will a fedora RPM work for centOS? |
| 18:08.11 | FLOSSrookie | ^ I moved to another distro... |
| 18:59.20 | Notify | 03BRL-CAD:carlmoore * 54885 brlcad/trunk/src/util/pixfilter.c: touchup so that pixfilter.c resembles, as close as possible, bwfilter.c . |
| 19:01.00 | brlcad | FLOSSrookie: did you try one of the rpms? |
| 19:01.06 | brlcad | rpm will tell you whether it works or not |
| 19:01.15 | brlcad | and if it doesn't, rpm can uninstall it ;) |
| 19:05.39 | FLOSSrookie | brlcad: I was concerned I might ruin the system by attempting such an install if they were not compatible. |
| 19:12.06 | FLOSSrookie | brlcad-7.22.0-0.x86_64 requires libstdc++.so.6(GLIBCXX_3.4.15)(64bit) |
| 19:12.06 | FLOSSrookie | brlcad-7.22.0-0.x86_64 requires libc.so.6(GLIBC_2.14)(64bit) |
| 19:13.07 | brlcad | what do you have installed? |
| 19:13.24 | FLOSSrookie | brlcad: Help me find out. |
| 19:13.28 | FLOSSrookie | What commands? |
| 19:19.13 | brlcad | oof, it's been a long time, I'd be running man rpm :) |
| 19:19.31 | brlcad | need to search what libstdc++ you have installed |
| 19:19.35 | brlcad | and libc |
| 19:19.43 | brlcad | if it's the same version, you're probably fine |
| 19:20.13 | brlcad | if you hang around long enough, jordi will be able to help you too |
| 19:26.13 | Notify | 03BRL-CAD:brlcad * 54886 brlcad/trunk/src/libbu/heap.c: implement a per-cpu data structure so that we can support multithreaded memory requests without pooching our data and without needing to acquire a performance-killer semaphore lock. restructure the bin data into a single struct container which should give better data coherence as well (but need to verify). |
| 19:32.22 | brlcad | ``Erik: notify web interface crashy crashy if I go to More graphs |
| 19:32.48 | brlcad | and applying a user filter didn't seem to do anything except give me a blue button with that user's name (didn't filter) |
| 19:33.55 | brlcad | and the main commit listing doesn't seem to be rendering right in IE (footer is just blue, truncates) |
| 20:19.03 | Notify | 03BRL-CAD:brlcad * 54887 (brlcad/trunk/include/bu.h brlcad/trunk/src/libbu/CMakeLists.txt brlcad/trunk/src/libbu/parallel.c): apply skriptkid's (akshay kashyap's) gci patch (http://www.google-melange.com/gci/task/view/google/gci2012/8070202) that implements cpu thread affinity. initial testing showed a HUGE impact on massively parallel systems with one 64-core ibm power7 system showing a benchmark performance |
| 20:19.05 | Notify | improvement around 50% (VGR increase from 120k to 180k). currently only implemented for pthread systems, but should be easily extended to other platforms and gives bu_parallel() some initial infrastructure so bu_parallel threads can know their CPU id. very cool. |
| 20:24.54 | Notify | 03BRL-CAD:brlcad * 54888 brlcad/trunk/src/libbu/affinity.c: ws style comment cleanup |
| 20:27.11 | Notify | 03BRL-CAD:brlcad * 54889 (brlcad/trunk/include/bu.h brlcad/trunk/src/libbu/affinity.c): move comments from source to header, remove unhelpful file comment, include common.h before system headers |
| 20:38.17 | Notify | 03BRL-CAD:brlcad * 54890 (brlcad/trunk/include/bu.h brlcad/trunk/src/libbu/CMakeLists.txt and 3 others): bu_set_affinity() doesn't need to be public API if we call it on all bu_parallel() threads for the user. it becomes an implementation detail. rename to parallel_set_affinity(), update accordingly to depublish, and add to old build. |
| 20:42.20 | Notify | 03BRL-CAD:brlcad * 54891 brlcad/trunk/src/libbu/parallel.c: clean up the array indexing, inconsistently unconventional |
| 20:42.48 | brlcad | only tested that on mac, so feel free to pound it on linux or other systems |
| 21:01.18 | ``Erik | brlcad: more graphs issue, izzit something like: The value "1 day" is not of type LOCAL-TIME:TIMESTAMP. |
| 21:01.57 | ``Erik | odd that the user filter doesn't work right, mebbe something was lost int he backing store migration :/ |
| 21:02.03 | ``Erik | and who the eff uses ie? O.o |
| 21:03.31 | ``Erik | ah, heh, lame, user/file filtering was all lost during backing store migration |
| 21:03.45 | ``Erik | !notify todo fix user/file filtering in project view |
| 21:03.45 | Notify | OK, added to https://elfga.com/notify/todo |
| 22:26.22 | *** join/#brlcad merzo (~merzo@92-33-132-95.pool.ukrtel.net) | |