| 00:53.21 | brlcad | starseeker: nod, just wondering -- often useful scavenge ideas from other devs that commit to hightly familiar code (such as you and scl's cmake logic) to see if there are ideas worth incorporating |
| 00:53.37 | brlcad | not necessarily code or exact names of things, but concepts that might be leveraged |
| 00:54.14 | brlcad | namely because if another dev needed it for some reason, likely someone else might too (and who best to review) |
| 00:55.30 | brlcad | jordi's need for mtune=generic is in a similar vein, actually, just a little less obvious |
| 00:57.47 | *** join/#brlcad merzo (~merzo@184-48-132-95.pool.ukrtel.net) | |
| 01:11.41 | CIA-128 | BRL-CAD: 03brlcad * r49481 10/brlcad/trunk/misc/CMake/CompilerFlags.cmake: |
| 01:11.41 | CIA-128 | BRL-CAD: add -mtune=generic to release builds in order to help ensure that -O3 |
| 01:11.41 | CIA-128 | BRL-CAD: compilations are still portable for a given target architecture. this may or |
| 01:11.41 | CIA-128 | BRL-CAD: may not be the issue reported in sf support request # 3464421 but give it a try. |
| 01:11.41 | CIA-128 | BRL-CAD: also, only set -std=gnu89 if we're a debug build. that way you can specify NONE |
| 01:11.42 | CIA-128 | BRL-CAD: for the build type and avoid both std flags. |
| 01:11.53 | brlcad | something for jordi to try when he gets back |
| 01:12.51 | brlcad | i'm still doubtful that is the issue, didn't think gcc defaulted to cpu-specific tuning at O3 or otherwise unless you set an march or mtune option |
| 01:13.08 | brlcad | given the crashes were in tcl, perhaps it's some flag in tcl's build logic.. |
| 02:17.47 | *** join/#brlcad merzo (~merzo@188-55-133-95.pool.ukrtel.net) | |
| 03:07.34 | starseeker | that's probably more likely - the tcl/tk logic is independent of our own |
| 03:07.44 | starseeker | (still our headache, of course...) |
| 03:08.40 | starseeker | I'll need to do another "apply lessons learned" pass over the src/other build systems at some point, and Tcl/Tk is usually where that comes up the most |
| 03:11.48 | *** join/#brlcad merzo (~merzo@110-63-133-95.pool.ukrtel.net) | |
| 04:23.39 | brlcad | a quick grep didn't uncover any mtune/march hints in tcl/tk/incr |
| 05:02.49 | *** join/#brlcad amit (3bb1cbb7@gateway/web/freenode/ip.59.177.203.183) | |
| 05:06.38 | Guest36931 | hi, I am an computer engineering 3rd year student, I wanted to take part in gsoc with your organisation. Can you please help regarding the steps i should take . |
| 05:11.31 | amit_12 | hi, I am an computer engineering 3rd year student, I wanted to take part in gsoc with your organisation. Can you please help regarding the steps i should take . |
| 05:38.39 | *** join/#brlcad jordisayol (~jordisayo@unaffiliated/jordisayol) | |
| 12:20.15 | jordisayol | brlcad: I see you added -mtune as flag |
| 12:21.41 | jordisayol | I made some test on my system, and it appear that -mtune=generic do not change anything on resulting object |
| 12:21.49 | jordisayol | $ test "$(gcc -m32 test.c -o test -O3 && md5sum test)" = "$(gcc -m32 test.c -o test -mtune=generic -O3 && md5sum test)" && echo "match" || echo "don't match" |
| 12:22.54 | jordisayol | return "match". same for g++ |
| 12:26.43 | jordisayol | "test.c" is a very simple printf("hello world!"); code. |
| 12:27.39 | jordisayol | s/don't match/mismatch/ |
| 13:23.25 | brlcad | jordisayol: yeah, I'm not convinced it'll do anything useful but it also shouldn't hurt |
| 13:24.14 | brlcad | your test isn't enough to exercise it, though .. you'd need something that would actually be optimized by -O3 in a potentially platform-specific way (math, lots of math and branching) |
| 13:24.14 | jordisayol | aha |
| 13:24.47 | brlcad | the real test is to make a new .deb compiled with mtune=generic and see if that user still gets an illegal instruction |
| 13:25.15 | brlcad | I suspect it'll still crash -- if it does, then the flag can be removed |
| 13:25.38 | jordisayol | aha, I'll ask to him if want to do this test |
| 13:26.02 | jordisayol | anyway, I'll remove the flags from deb building process |
| 13:26.13 | brlcad | nods |
| 13:27.38 | jordisayol | sorry, nods? :-/ |
| 13:28.09 | brlcad | I'm nodding my head :) |
| 13:28.18 | brlcad | it means I understand or I agree or both :) |
| 13:28.46 | jordisayol | ok, many thanks for teach me :-) |
| 13:28.53 | brlcad | no problem |
| 14:01.56 | CIA-128 | BRL-CAD: 03starseeker * r49482 10/brlcad/trunk/src/other/libpng/ (34 files in 5 dirs): Update libpng to 1.5.9 |
| 14:04.47 | CIA-128 | BRL-CAD: 03starseeker * r49483 10/brlcad/trunk/src/other/libpng/Makefile.am: re-add the changes to Makefile.am from 46025 |
| 14:07.46 | CIA-128 | BRL-CAD: 03starseeker * r49484 10/brlcad/trunk/src/other/libpng/CMakeLists.txt: Go with 'lib' for all instances of CMAKE_LIBRARY_OUTPUT_DIRECTORY (r49115) |
| 14:09.03 | CIA-128 | BRL-CAD: 03starseeker * r49485 10/brlcad/trunk/src/other/libpng/projects/vstudio/pngstest/: pngstest is still empty (r49239) |
| 15:00.50 | CIA-128 | BRL-CAD: 03jordisayol * r49486 10/brlcad/trunk/misc/debian/rules: remove unneeded flags on deb building process. |
| 15:08.20 | brlcad | starseeker: awesome, I was just thinking about that update yesterday .. there's a CVE out on that one |
| 15:08.33 | brlcad | goes to get a rental |
| 15:52.52 | *** join/#brlcad Stattrav (~Stattrav@unaffiliated/stattrav) | |
| 17:55.46 | CIA-128 | BRL-CAD: 03starseeker * r49487 10/brlcad/trunk/src/librtserver/CMakeLists.txt: Get librtserver closer to building properly. |
| 17:57.32 | CIA-128 | BRL-CAD: 03starseeker * r49488 10/brlcad/branches/STABLE/src/librtserver/CMakeLists.txt: Add the librtserver build tweaks to STABLE |
| 18:08.58 | CIA-128 | BRL-CAD: 03starseeker * r49489 10/brlcad/branches/STABLE/src/librtserver/CMakeLists.txt: Ah, right - too early for that version of the INCLUDE_DIRS macro |
| 18:10.25 | CIA-128 | BRL-CAD: 03starseeker * r49490 10/brlcad/branches/STABLE/src/librtserver/CMakeLists.txt: Fix form, not just name... |
| 19:07.41 | *** join/#brlcad merzo (~merzo@110-63-133-95.pool.ukrtel.net) | |
| 20:02.27 | *** join/#brlcad merzo (~merzo@87-52-132-95.pool.ukrtel.net) | |
| 20:05.18 | CIA-128 | BRL-CAD: 03brlcad * r49491 10/brlcad/trunk/TODO: svg is in the same boat as pdf conversion export |
| 20:57.20 | CIA-128 | BRL-CAD: 03brlcad * r49492 10/brlcad/trunk/TODO: benchmark needs to include compiler details too. |
| 22:27.17 | CIA-128 | BRL-CAD: 03bob1961 * r49493 10/brlcad/trunk/src/tclscripts/lib/TkTable.tcl: Update cadwidgets::TkTable::handleKey to potentially call the dataCallback when a cell in column 0 receives a keypress. |
| 22:54.34 | CIA-128 | BRL-CAD: 03brlcad * r49494 10/brlcad/trunk/BUGS: |
| 22:54.34 | CIA-128 | BRL-CAD: the yacc parsers are failing on non-bison platforms due to differences in |
| 22:54.34 | CIA-128 | BRL-CAD: behavior. it'll be at least a few months before all are converted to lemon, so |
| 22:54.34 | CIA-128 | BRL-CAD: note the bug in the meantime. the fix looks pretty simple (FindYACC.cmake), but |
| 22:54.34 | CIA-128 | BRL-CAD: needs testing. |
| 23:20.28 | *** join/#brlcad piksi (piksi@pi-xi.net) | |