| 01:29.37 | *** join/#brlcad ymffhufurpxdsuox (~armin@dslb-092-074-238-107.092.074.pools.vodafone-ip.de) | |
| 01:33.41 | brlcad | neat idea ... http://tldr.sh |
| 01:55.17 | brlcad | starseeker: gdal compile errors on -mavx and -mssse3 not being valid compiler flags, presumably need to test for them |
| 02:34.59 | brlcad | think I might have sorted out a relatively easy fix |
| 03:32.12 | Stragus | Why would -mavx and -mssse3 be invalid compiler flags? Both GCC and clang accept them... |
| 03:42.16 | brlcad | either hardware or compiler version |
| 03:42.39 | brlcad | it was on a 100-core Power (big endian) system |
| 03:52.04 | Stragus | Ah yes, probably invalid flags on non-x86 |
| 04:15.27 | *** join/#brlcad virtualmachine (~virtual@c-69-243-71-53.hsd1.md.comcast.net) | |
| 04:15.44 | virtualmachine | testing |
| 04:19.24 | *** join/#brlcad virtualmachine (~virtual@c-69-243-71-53.hsd1.md.comcast.net) | |
| 04:20.06 | *** join/#brlcad virtualmachine (~virtual@c-69-243-71-53.hsd1.md.comcast.net) | |
| 04:22.15 | *** join/#brlcad caxuser (~caxuser@c-69-243-71-53.hsd1.md.comcast.net) | |
| 04:22.49 | *** join/#brlcad caxuser_ (~caxuser@c-69-243-71-53.hsd1.md.comcast.net) | |
| 05:16.06 | Notify | 03BRL-CAD:brlcad * 70436 brlcad/trunk/misc/CMake/CMakeLists.txt: add some logic to detect whether sse/avx/simd support is available. credit pytorch folks where found, though unconfirmed that they are the original authors. modified to not perform tests simply by finding the package. currently used by gdal build logic, should be extended to cad. improvement over existing sse cmake detection logic. |
| 05:18.00 | Notify | 03BRL-CAD:brlcad * 70437 brlcad/trunk/src/libged/simulate/bullet/CMakeLists.txt: fix distcheck file inclusions, use our BRLCAD_ADDLIB wrapper since this isn't going upstream (and is in a BRLCAD_ENABLE_BULLET specific block). |
| 05:18.33 | Notify | 03BRL-CAD:brlcad * 70438 brlcad/trunk/src/other/gdal/port/cpl_cpu_features.cpp: fix return type |
| 05:19.51 | Notify | 03BRL-CAD:brlcad * 70439 (brlcad/trunk/src/other/gdal/CMakeLists.txt brlcad/trunk/src/other/gdal/alg/CMakeLists.txt brlcad/trunk/src/other/gdal/gcore/CMakeLists.txt): need to conditionally use the -mavx and -mssse3 flags; and conditionally set whether we have the respective simd instructions sets. fixed bug in flag definition being set (HAVE_SSSE_AT... was wrong) |
| 06:23.49 | *** join/#brlcad gabbar1947 (uid205515@gateway/web/irccloud.com/x-kjovgssabsecjhtn) | |
| 06:26.11 | Notify | 03BRL-CAD:brlcad * 70440 brlcad/trunk/src/libnmg/inter.c: fix array subscript is above array bounds. looks like the intent is to iterate through the edge uses to find and join adjacent ones, which means we don't need to go to the end of the list. we can stop 1 shy (neu-2) with the inner loop going from next to end. |
| 06:44.07 | Notify | 03BRL-CAD:brlcad * 70441 brlcad/trunk/src/other/libpng/pngpriv.h: doesn't look like the vsx code is actually here. maybe files that didn't make it from upstream sync or were inadvertently removed. should fix enableall build on powerpc. |
| 09:24.30 | *** join/#brlcad merzo (~merzo@185.39.197.205) | |
| 10:39.41 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 10:47.28 | *** join/#brlcad AndChat264089 (~AndChat26@2405:204:1300:a46c:99a8:9bc2:6996:f322) | |
| 10:50.32 | AndChat264089 | Hello,sorry if this question seems quite basic.I do quite a bit of competetive programming on sites like codechef.For some time Ive been looking forward to open source and google code in 2017.I know c++ to a good extent including the data structures and dynamic programming,but I cant figure out where to begin in open source.All the projects and issues/bugs I see seem completely alien to me and I can't understand how my c++ s |
| 11:13.14 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 11:23.27 | Stragus | AndChat264089: message was cut at "how my c++ s" |
| 11:23.51 | Stragus | Usually, you need to spend some time getting familiar with a project and its source code before you can submit patches |
| 11:32.16 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 11:42.02 | AndChat264089 | *contd msg how my c++ skills can help solve them.What am I missing here?Is there something special to learnt, other than the language itself, for open source development? |
| 11:42.42 | AndChat264089 | Okay,do I need to learn something else apart from the language? I've heard talks of frameworks and APIs? |
| 11:51.31 | Stragus | There's a ton of APIs and libraries you could learn and use, but that all depends on the projects involved |
| 11:51.51 | Stragus | For example, if you do 3D graphics, you might want to learn OpenGL |
| 11:58.18 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 12:15.47 | AndChat264089 | Ok, thanks. Are there any APIs that I should learn for brlcad? |
| 12:17.30 | Stragus | What parts would you like to work on? There's demand for OpenCL to trace rays on GPUs, for example |
| 12:18.50 | AndChat264089 | Ok, guess I'll have to research a bit first then |
| 12:23.51 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 12:30.10 | *** join/#brlcad merzo (~merzo@185.39.197.205) | |
| 12:52.22 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 12:52.50 | starseeker | brlcad: thanks |
| 13:13.59 | *** join/#brlcad yorik (~yorik@2804:431:f721:544a:290:f5ff:fedc:3bb2) | |
| 13:25.22 | Notify | 03BRL-CAD:Jessestichter * 0 /wiki/User:Jessestichter: |
| 14:33.45 | Notify | 03BRL-CAD:starseeker * 70442 brlcad/trunk/src/libgcv/plugins/gdal/gdal.cpp: Just use the transform we already get (fGeoT) rather than getting another. |
| 14:35.29 | Notify | 03BRL-CAD:starseeker * 70443 brlcad/trunk/src/util/pixcrop.c: fix shadowing |
| 14:37.43 | Notify | 03BRL-CAD:starseeker * 70444 brlcad/trunk/src/util/bwcrop.c: same issue in bwcrop |
| 14:43.14 | brlcad | starseeker: sure, let me know if I botched it anywhere .. |
| 14:43.22 | Notify | 03BRL-CAD:starseeker * 70445 brlcad/trunk/misc/CMake/FindSSE.cmake: clear Policy CMP0054 warning, downcase |
| 15:03.17 | Notify | 03BRL-CAD:starseeker * 70446 brlcad/trunk/CMakeLists.txt: Force bullet off until it's fixed |
| 15:04.51 | brlcad | fwiw, bullet build was back to working on bsd and at least one linux box |
| 15:13.32 | Notify | 03BRL-CAD Wiki:Apoorv.jain25 * 0 /wiki/User:Apoorv.jain25: |
| 15:16.53 | Notify | 03BRL-CAD:brlcad * 70447 brlcad/trunk/src/libnmg/inter.c: underlying issue being warned is that someone set an upper limit without checking if it got exceeded. |
| 15:21.50 | starseeker | brlcad: getting some include errors here, and I'm 99.9999% sure the Windows build is totally fried |
| 15:22.12 | starseeker | even then, dropping ktank on the dsp example doesn't work, so either way we're not ready for prime time |
| 15:25.04 | brlcad | yeah, hadn't even gotten to windows yet because of other build failures being worked out |
| 15:25.35 | starseeker | nods |
| 15:26.14 | starseeker | i've got to say in some ways the bullet source code (version 2.x anyway) is rather annoying |
| 15:26.44 | brlcad | does libged/simulate/bullet build successfully? |
| 15:26.47 | starseeker | supposes it would be an insane amount of work to recast the logic into our data types, but sometimes it's tempting to try |
| 15:26.54 | starseeker | not for me currently |
| 15:27.02 | starseeker | can't find headers |
| 15:27.17 | brlcad | huh |
| 15:27.25 | starseeker | brlcad: it's probably not worth fooling with with GCI about to kick off... |
| 15:27.49 | brlcad | makes it a GCI task |
| 15:27.52 | starseeker | if it worked I'd feel differently, but there's still a fair bit to do functionality wise |
| 15:28.46 | starseeker | I have no clue what to do about the Windows build if we're not going to build bullet as part of libged - I tried DLL, I tried static, I tried adding exports to support DLL... all were rabbit holes |
| 15:29.50 | starseeker | upstream apparently expects people to build it static and use it that way |
| 15:30.02 | brlcad | that's curious though because any of those should work .. so is something wrong in BRLCAD_ADDLIB? |
| 15:30.25 | starseeker | shakes his head - possibly, but I get the sense that bullet isn't designed/coded for DLL linking |
| 15:30.48 | starseeker | I don't recall what went wrong with the static option - that could be me |
| 15:30.52 | brlcad | well yeah, I get why that one would fail if there's not a symbol file |
| 15:34.22 | brlcad | static doesn't require import/export mess so that's why I put it on the flag list |
| 15:34.38 | starseeker | brlcad: we may be worse off than most because of the level at which we need to interface with bullet (feeding ray data to it...) |
| 15:35.04 | starseeker | brlcad: worth trying - I did have a go with it, but I may have done something wrong |
| 15:36.05 | brlcad | the big studios that hooked into bullet use it far harder and more complex than we're using :) |
| 15:36.17 | brlcad | hm, looks like we're a little behind versioning too |
| 15:36.26 | brlcad | another possibility |
| 15:37.51 | starseeker | brlcad: they're also probably not shy about shoehorning it in any old way... |
| 15:40.42 | starseeker | wonders if https://swiftshader.googlesource.com/SwiftShader might make a good candidate/alternative to Mesa to guarantee an OpenGL rendering capability in the event of driver problems... |
| 15:41.55 | brlcad | well, I think we can sort it out easily enough either way .. it seems to be close |
| 15:42.15 | brlcad | is src/libged/simulate/bullet/CMakeLists.txt file yours or from the bullet guys? |
| 15:46.19 | starseeker | I think that one's me |
| 15:47.14 | starseeker | IIRC, I pulled some bits from their other files |
| 15:49.03 | starseeker | I was trying to at least avoid issues linking bullet's separate libraries, so I build one big library with all of the parts we need |
| 15:54.19 | brlcad | nods, it's simple enough |
| 15:56.24 | starseeker | I suppose I shouldn't ask this, but I wonder if we could mod qu3e more easily to use rays and be "good enough" |
| 17:26.51 | brlcad | relevant: http://www.adrianboeing.com/pal/papers/p281-boeing.pdf |
| 17:35.35 | Notify | 03BRL-CAD:brlcad * 70448 brlcad/trunk/CMakeLists.txt: running into a variety of other link-time-optimization bugs that can't be worked around. avoid LTO on busted versions. not easily detected in a feature test (looking not even possible outside of tailoring to compiler-specific bugs, which seems overkill). |
| 17:44.11 | *** join/#brlcad merzo (~merzo@78-26-133-95.pool.ukrtel.net) | |
| 19:31.52 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 20:23.31 | *** join/#brlcad merzo (~merzo@78-26-133-95.pool.ukrtel.net) | |
| 20:26.55 | *** join/#brlcad merzo (~merzo@78-26-133-95.pool.ukrtel.net) | |
| 20:34.26 | *** join/#brlcad teepee (~teepee@unaffiliated/teepee) | |
| 20:42.44 | *** join/#brlcad merzo (~merzo@78-26-133-95.pool.ukrtel.net) | |
| 21:11.52 | *** join/#brlcad merzo (~merzo@238-12-133-95.pool.ukrtel.net) | |
| 21:53.08 | *** join/#brlcad merzo (~merzo@238-12-133-95.pool.ukrtel.net) | |