| 00:22.04 | *** join/#brlcad infobot (ibot@rikers.org) | |
| 00:22.04 | *** topic/#brlcad is GSoC students: if you have a question, ask and wait for an answer ... responses may take minutes or hours. Ask and WAIT. ;) | |
| 00:27.52 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 01:27.28 | *** join/#brlcad DaRock (~Thunderbi@mail.unitedinsong.com.au) | |
| 03:03.35 | brlcad | good article, lots of solid best practice points: https://opensource.com/article/17/5/30-best-practices-software-development-and-testing?sc_cid=70160000000xYJ9AAM |
| 03:03.36 | gcibot_ | [ 30 best practices for software development and testing | Opensource.com ] |
| 03:11.52 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 03:12.08 | Stragus | bookmarks that just half-way through reading |
| 03:14.01 | Stragus | can't agree with #17 though |
| 03:16.16 | brlcad | yeah, I cringed at that one a bit |
| 03:30.53 | Stragus | I'm unsure how common it is, but I have found invaluable this little trick of being able to turn on some #define and produce very verbose logs on any and all problems or sub-problems |
| 03:31.28 | Stragus | I pretty much always do it now, even before any bug is found or suspected, just to be able to quickly inspect the internal flow if ever needed |
| 03:40.57 | Notify | 03BRL-CAD:brlcad * 69867 brlcad/trunk/src/other/tcl/CMakeLists.txt: define the right thing for the AvailabilityMacros.h test |
| 03:42.52 | Notify | 03BRL-CAD:brlcad * 69868 brlcad/trunk/src/other/tcl/CMakeLists.txt: these have been wrong for some time, define the right symbols |
| 03:44.17 | brlcad | i do that for standalone and some unit tests, but rarely for production code |
| 03:44.39 | brlcad | but brl-cad employs that pattern very pervasively for debug logging |
| 03:50.15 | Stragus | Not for production code? I see |
| 03:50.52 | Stragus | It's also very useful to validate usage of some API. "I'm doing X and it doesn't work!" ~ I ask the log and I can see everything, including bad input/usage |
| 03:52.29 | brlcad | good unit tests can be useful for that too, one of the points they make |
| 03:52.46 | brlcad | they become defacto examples of usage |
| 03:54.05 | Stragus | Yes well, I still had people try to trace rays with vectors of NaNs, triangulate overlapping vertices, smooth lines with an interval of zero |
| 03:54.58 | brlcad | good unit tests always check boundary conditions including nan and inf ;) |
| 03:55.55 | Stragus | I would argue that there are valid performance reasons to intentionally break/fail on non-sense input :) |
| 03:58.00 | brlcad | fast is useless if it doesn't handle their data, and geometry data is pervasively dirty, invalid, non-sense |
| 03:58.47 | brlcad | in my experience, it's above 90% of production models have at least some invalid geometry |
| 03:58.50 | Stragus | Yes, I have realized that recently once again... OpenStreetMaps data is full of non-sense. Contours that overlap themselves, you name it |
| 03:59.45 | brlcad | academic data sets are about the only thing I run into that tends to be clean |
| 03:59.55 | brlcad | unless it's an academic data set about unclean data ;) |
| 04:01.10 | Stragus | I'm actually annoyed by designing clever, robust algorithms that work perfectly with good input, including handling of limited floating point precision. Then people feed it bad input and complain |
| 04:01.22 | Stragus | But that's just what the data is |
| 04:02.20 | brlcad | so write a clever robust input filter that does a good job cleaning up bad geometry ;) |
| 04:02.39 | Stragus | Yes, I am learning that, once again... :) |
| 04:48.58 | Notify | 03BRL-CAD:brlcad * 69869 (brlcad/trunk/src/other/tcl/CMake/tcl.cmake brlcad/trunk/src/other/tcl/CMakeLists.txt): bug has probably been in here since the beginning, was setting HAVE_HAVE_AVAILABILITYMACROS_H and not checking a couple other headers. change the funcs to just take the var we want to have set (helps when searching for it) instead of appeanding HAVE_ automatically. avoids potential inconsistencies down |
| 04:49.00 | Notify | the road too. |
| 04:49.02 | Notify | ... |
| 04:52.49 | Notify | 03BRL-CAD:brlcad * 69870 brlcad/trunk/src/other/tcl/CMake/tcl.cmake: closer to matching, we don't need the NO and SIZEOF defines |
| 05:09.32 | Notify | 03BRL-CAD:brlcad * 69871 brlcad/trunk/src/other/tcl/CMake/tcl.cmake: still need to make the tcl.m4 file, which manually defines the NO_ vars so we should too. |
| 05:18.04 | *** join/#brlcad DaRock (~Thunderbi@mail.unitedinsong.com.au) | |
| 06:42.14 | *** join/#brlcad teepee_ (~teepee@unaffiliated/teepee) | |
| 07:06.16 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 07:44.56 | *** join/#brlcad Caterpillar (~caterpill@unaffiliated/caterpillar) | |
| 07:59.58 | *** join/#brlcad KimK (~Kim__@2600:8803:7a81:7400:8045:43c4:51ee:69e9) | |
| 08:43.54 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 08:56.25 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 09:03.01 | *** join/#brlcad merzo (~merzo@94.45.58.139) | |
| 09:51.00 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 10:11.05 | *** join/#brlcad teepee (~teepee@p200300D4DBCCB86DBA27EBFFFEAEED55.dip0.t-ipconnect.de) | |
| 10:11.05 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 10:53.35 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 11:02.17 | Notify | 03BRL-CAD:Amritpal singh * 10005 /wiki/User:Amritpal_singh/GSoC17/logs: |
| 11:33.07 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 11:50.52 | Notify | 03BRL-CAD:brlcad * 69872 brlcad/trunk/src/other/tcl/CMake/tcl.cmake: fix typo, dlfcn |
| 12:03.06 | *** join/#brlcad d_rossberg (~rossberg@104.225.5.10) | |
| 12:40.58 | Notify | 03BRL-CAD:brlcad * 69873 (brlcad/trunk/src/other/tcl/CMake/tcl.cmake brlcad/trunk/src/other/tcl/CMakeLists.txt): set MODULE_SCOPE.... also, add support for setting TCL_WIDE_INT_IS_LONG=1 mimicing the essential tcl.m4 bits while assuming long long (since we do). with this, there are no longer defines missing when compared with the autotools tcl build except for HAVE_CAST_TO_UNION which is a silly bit for windows |
| 12:41.00 | Notify | (and unnecessary). that is, they match (superset). |
| 12:41.02 | Notify | ... |
| 12:51.39 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 12:51.58 | *** join/#brlcad teepee] (bc5c2134@gateway/web/freenode/ip.188.92.33.52) | |
| 13:20.29 | Notify | 03BRL-CAD:starseeker * 69874 (brlcad/branches/tcltk86/src/other/tcl/CMake/tcl.cmake brlcad/branches/tcltk86/src/other/tcl/CMakeLists.txt): pull at least some of the fixes from the 8.5 tcl build to the new 8.6 logic, but undoubtedly we'll have to redo the comparison with the autotools outputs when this gets read to go live... |
| 13:24.25 | starseeker | brlcad: intersting! (clang-tidy) |
| 13:25.14 | starseeker | I was vaguely aware of it, but I think I had filed it in the "to-look-at-after-we-finish-clearing-clang-static-analysis-warnings" bin |
| 13:25.52 | starseeker | the "write your own check" piece sounds very interesting |
| 13:39.46 | *** join/#brlcad yorik (~yorik@2804:431:f720:1720:290:f5ff:fedc:3bb2) | |
| 14:25.47 | Notify | 03BRL-CAD:starseeker * 69875 (brlcad/trunk/src/librt/cache.c brlcad/trunk/src/librt/tree.c): Try to tweak the rt_cache logic so that failing to create/open a cache isn't fatal. We have a fallback behavior - just prep as we always have previously - which is preferable to catastrophic failure. |
| 16:34.29 | *** join/#brlcad gabbar1947 (uid205515@gateway/web/irccloud.com/x-yjzapdkpuknoeoot) | |
| 16:36.31 | *** join/#brlcad Amritpal (~amritpal@124.253.104.50) | |
| 17:42.55 | Notify | 03BRL-CAD:brlcad * 69876 (brlcad/trunk/src/other/tk/CMake/tcl.cmake brlcad/trunk/src/other/tk/CMakeLists.txt): sync with tcl, lots of cleanup to more closely match tcl/tk autotools build system |
| 18:15.03 | Notify | 03BRL-CAD:brlcad * 69878 brlcad/trunk/src/other/tk/CMakeLists.txt: match tk's version defines the same, test for the posix headers so we get the limits.h check |
| 18:16.36 | Notify | 03BRL-CAD:brlcad * 69877 brlcad/trunk/src/other/tcl/CMakeLists.txt: tweak to match what tcl uses |
| 18:24.01 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 18:43.56 | Notify | 03BRL-CAD:starseeker * 69879 brlcad/trunk/TODO: workaround for osx drawing problem was nulling out color - fixed |
| 18:50.13 | Notify | 03BRL-CAD:Amritpal singh * 10006 /wiki/User:Amritpal_singh/GSoC17/logs: /* Coding Period */ |
| 18:52.34 | Notify | 03BRL-CAD:Marco-domingues * 10007 /wiki/User:Marco-domingues/GSoC17/Log: |
| 18:57.35 | Notify | 03BRL-CAD:Marco-domingues * 0 /wiki/File:Csg_simple_example.png: CSG test scene with the intersection of 2 spheres |
| 19:02.24 | Notify | 03BRL-CAD:Marco-domingues * 10009 /wiki/User:Marco-domingues/GSoC17/Log: /* Week 1: 30 May - 2 June */ |
| 19:06.49 | Notify | 03BRL-CAD:Gabbar1947 * 10010 /wiki/Google_Summer_of_Code: /* BRL-CAD participation in GSoC */ |
| 19:11.21 | Notify | 03BRL-CAD:starseeker * 69880 (brlcad/trunk/CMakeLists.txt brlcad/trunk/TODO): Add PSAPI_VERSION definition - needs testing. |
| 19:12.01 | Notify | 03BRL-CAD:starseeker * 69881 brlcad/trunk/TODO: leaving astyle where it is for now due to increased compiler requirements. reasons are documented, can be revisited later. |
| 19:13.46 | Notify | 03BRL-CAD:starseeker * 69882 brlcad/trunk/TODO: If we still want these at all, it's going to be a while before we get to them... |
| 19:15.10 | Notify | 03BRL-CAD:Marco-domingues * 10011 /wiki/User:Marco-domingues/GSoC17/Log: /* Week 1: 30 May - 2 June */ |
| 19:25.20 | Notify | 03BRL-CAD:brlcad * 69883 brlcad/trunk/src/other/tk/CMakeLists.txt: couple more symbols to sync. check for pw_gecos and an Xkb symbol. might not be the most robust test, but is working and close to the autotools version minus the clunky way we're testing for the symbol. |
| 19:33.38 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 19:40.00 | *** join/#brlcad Amritpal (~amritpal@124.253.104.50) | |
| 19:41.23 | Notify | 03BRL-CAD:Amritpal singh * 10012 /wiki/User:Amritpal_singh/GSoC17/logs: /* Coding Period */ |
| 19:52.43 | Notify | 03BRL-CAD:brlcad * 69884 brlcad/trunk/src/other/tk/CMakeLists.txt: and this closes the loop, now matching defining everything in tk's list (we define a superset) except again for the silly HAVE_CAST_TO_UNION. |
| 21:15.07 | brlcad | that gets the builds in sync so we can update to the lastest 8.5 release, which doesn't have the mac segfault |
| 22:28.37 | *** join/#brlcad Caterpillar (~caterpill@unaffiliated/caterpillar) | |
| 22:31.53 | *** join/#brlcad Ch3ck_ (~Ch3ck@104.225.5.10) | |
| 22:40.25 | *** join/#brlcad LordOfBikes (~armin@dslb-088-066-139-129.088.066.pools.vodafone-ip.de) | |
| 23:37.05 | *** join/#brlcad teepee_ (~teepee@unaffiliated/teepee) | |