| 00:00.02 | starseeker | but it did avoid introducing the observed overlap |
| 00:00.28 | starseeker | isn't particularly attached to r67850 if you'd prefer to revert... |
| 00:02.03 | starseeker | brlcad: anything I can to to help with release prep? |
| 00:02.15 | starseeker | is doing another test build on Windows... |
| 00:16.01 | *** join/#brlcad Mathnerd314 (~quassel@supertux/Mathnerd314) | |
| 00:23.34 | brlcad | don't care much about r67850, think it doesn't do much overall |
| 00:24.51 | brlcad | clamping introduces and fixes problems because it's essentially making corrections so far out on the decimal that are introduced simply by loading geoemtry into floating point registers |
| 00:25.24 | brlcad | so not clamping will obviously not introduce but also not fix other error being introduced |
| 00:26.26 | brlcad | e.g., consider a modeler that uses the in command to make some geometry at x=52, and it gets written out as 52.0000001000100301235345 ... preserving that without clamping is preserving error we introduced |
| 00:38.15 | Stragus | So someone is complaining because converting from binary to ascii and back reduces precision? |
| 00:38.17 | Stragus | o.O |
| 00:45.14 | Stragus | Although, if you print enough digits, I believe glibc guarantees you'll get the same binary representation back |
| 00:46.03 | starseeker | Stragus: not directly - they're complaining because an overlap manifested with gqa that didn't manifest before the conversion - basically, it was "right on the edge" and the g2asc->asc2g round trip nudged it just enough |
| 00:46.55 | starseeker | that's why I left all of the calls to the matrix encoder as clamping except the low level one in the tree walker |
| 00:47.17 | starseeker | the idea being "what's there is there, but what we're editing/changing gets clamped." |
| 00:49.09 | *** join/#brlcad byckxtpgmsptffqr (~armin@dslc-082-083-186-230.pools.arcor-ip.net) | |
| 01:24.35 | brlcad | starseeker: I think it's for the better, leave it that way |
| 01:25.05 | brlcad | only downside is the API inconsistency with the other routines that do the same thing for non-matrix data |
| 02:46.18 | starseeker | nods - I can update those tomorrow to be consistent if you like |
| 03:27.30 | brlcad | it's undocumented API, so can update or leave until it for distillation removal review |
| 03:31.19 | Notify | 03BRL-CAD:brlcad * 67851 brlcad/trunk/include/bu/ptbl.h: revisit work back in r57743 and r57742 where we try to avoid ptbl crashes when referencing NULL, trying to give our bomb checks a chance to catch them more gracefully. since pointers could be data on the stack, we need to trick the compiler a bit to not warn about the test always being true (it's the ones that are not that we care more about). we do this by |
| 03:31.21 | Notify | forcing a conversion to int first. |
| 03:31.23 | Notify | ... |
| 03:48.02 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 04:08.03 | Notify | 03BRL-CAD:brlcad * 67852 brlcad/trunk/src/libbu/ptbl.c: allow truncating to any size less than the initialized allocation. this lets us to set a size in both directions. |
| 04:09.54 | Notify | 03BRL-CAD:brlcad * 67853 brlcad/trunk/src/librt/prep.c: use bu_ptbl_trunc now that it will set the end marker to decrease or increase, but to some truncation at or under the initialized size. |
| 04:14.48 | Notify | 03BRL-CAD:brlcad * 67854 brlcad/trunk/src/librt/primitives/submodel/submodel.c: eliminate the last instance where we were accessing ->end merely to expand the size for null pointers. instead, call bu_ptbl_trunc to set it. |
| 04:17.03 | Notify | 03BRL-CAD:brlcad * 67855 brlcad/trunk/src/libbu/ptbl.c: make bu_ptbl_trunc take a size_t instead of an int for better intent-based comparison and warning-free compiles |
| 04:26.33 | *** join/#brlcad poxip (~poxip@unaffiliated/mrpoxipol) | |
| 05:27.22 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 06:57.52 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 06:58.12 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 07:37.53 | *** join/#brlcad teepee` (bc5c2134@gateway/web/freenode/ip.188.92.33.52) | |
| 08:07.40 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 08:10.39 | *** join/#brlcad catchchaos (6ad8958b@gateway/web/freenode/ip.106.216.149.139) | |
| 08:19.36 | *** part/#brlcad catchchaos (6ad8958b@gateway/web/freenode/ip.106.216.149.139) | |
| 08:57.37 | *** join/#brlcad criado (~criado@edurd1.unican.es) | |
| 09:09.40 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 09:24.33 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 10:39.05 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 12:39.33 | *** join/#brlcad yorik (~yorik@177.189.140.33) | |
| 12:42.59 | Notify | 03BRL-CAD:starseeker * 67856 (brlcad/trunk/include/bn/str.h brlcad/trunk/src/libbn/str.c and 19 others): Add a clamp flag to the other bn_encode functions for consistency. |
| 13:07.14 | *** join/#brlcad amarjeet (~Amarjeet@101.214.126.143) | |
| 13:11.22 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 13:54.11 | *** join/#brlcad Mathnerd314 (~quassel@supertux/Mathnerd314) | |
| 14:30.43 | *** join/#brlcad tandoorichick (b64b2d01@gateway/web/freenode/ip.182.75.45.1) | |
| 14:39.44 | Notify | 03BRL-CAD:starseeker * 67857 brlcad/trunk/src/libbu/dirent.c: Use bu_strcmp to keep regress-repository happy |
| 15:01.45 | *** join/#brlcad sniok (~sniok@89.252.29.238) | |
| 15:19.04 | Notify | 03BRL-CAD:brlcad * 67858 brlcad/trunk/include/bu/ptbl.h: make bu_ptbl_trunc take a size_t instead of an int for better intent-based comparison and warning-free compiles |
| 15:19.10 | brlcad | oops, missed committing a file |
| 15:33.02 | brlcad | starseeker: testing now, but I think that was all that was missing? |
| 15:45.05 | *** join/#brlcad amarjeet (~Amarjeet@101.214.30.65) | |
| 15:50.20 | brlcad | ah, now I'm getting it .. some reason had to wipe out the build dir |
| 16:07.44 | Notify | 03BRL-CAD:starseeker * 67859 brlcad/trunk/AUTHORS: Add commit name for D. Remenak |
| 16:17.11 | Notify | 03BRL-CAD:starseeker * 67860 brlcad/trunk/AUTHORS: Add Doug's commit name |
| 16:19.11 | Notify | 03BRL-CAD:starseeker * 67861 brlcad/trunk/AUTHORS: Associate Issac's commit name |
| 16:21.02 | Notify | 03BRL-CAD:starseeker * 67862 brlcad/trunk/AUTHORS: associate Peter's commit name |
| 17:12.48 | *** join/#brlcad tandoorichick (3d0c28b1@gateway/web/freenode/session) | |
| 17:16.28 | *** join/#brlcad shubham (71c1890f@gateway/web/freenode/session) | |
| 18:41.56 | *** join/#brlcad shubham (71c1890f@gateway/web/freenode/ip.113.193.137.15) | |
| 19:02.15 | *** join/#brlcad criado (~criado@47.61.245.238) | |
| 19:14.46 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 19:55.56 | Notify | 03BRL-CAD:brlcad * 67863 brlcad/trunk/src/librt/bool.c: some compilers, ither being clever or erroneous, are noticing the potential NULL back from the ternary operator and warning. stashing into a pointer first does the trick. |
| 20:27.57 | brlcad | starseeker: fixed |
| 20:39.04 | starseeker | brlcad: awesome, thanks! |
| 20:50.23 | *** join/#brlcad kintel (~kintel@unaffiliated/kintel) | |
| 22:43.53 | *** join/#brlcad skat00sh (uid103741@gateway/web/irccloud.com/x-lxwuieqpgnkjpugd) | |