--- Log opened Wed Aug 14 00:00:14 2013 | ||
@wiking | sonney2k: no not really as src could be anywhere | 00:01 |
---|---|---|
@wiking | coz of out of source compilation i cannot assume that .. is the root of shogun | 00:02 |
@wiking | sonney2k: but i'll look into it how we can get around this | 00:02 |
@sonney2k | wiking, but you have $ROOT/src right? | 00:03 |
@wiking | i have src root yes | 00:03 |
@wiking | ${CMAKE_SOURCE_DIR} is the root of the source | 00:04 |
@sonney2k | wiking, that would work then for the include | 00:04 |
@wiking | but we do include that | 00:05 |
@wiking | but i guess the order here metters | 00:05 |
@sonney2k | wiking, yes first thing is taken first | 00:08 |
@wiking | mmm | 00:09 |
shogun-buildbot | build #1770 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1770 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:09 |
@iglesiasg | wiking, hey! I have an issue, I am not sure if it can be caused by some configuration in cmake, or something I am missing about it | 00:21 |
@iglesiasg | wiking, I compile python modular and when I try to import the LMNN class with from shogun.Metric import LMNN python crashes | 00:22 |
@iglesiasg | import shogun.Metric works fine | 00:22 |
@iglesiasg | other imports are fine too, e.g. from shogun.Classifier import KNN | 00:22 |
@sonney2k | iglesiasg, backtrace / valgrind! | 00:24 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has joined #shogun | 00:24 | |
travis-ci | [travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10171993 | 00:24 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has left #shogun [] | 00:24 | |
@iglesiasg | sonney2k, Traceback (most recent call last): | 00:24 |
@iglesiasg | File "metric_lmnn_modular.py", line 72, in <module> | 00:24 |
@iglesiasg | metric_lmnn_modular() | 00:24 |
@iglesiasg | File "metric_lmnn_modular.py", line 10, in metric_lmnn_modular | 00:24 |
@iglesiasg | from shogun.Metric import LMNN | 00:24 |
@iglesiasg | ImportError: cannot import name LMNN | 00:24 |
@iglesiasg | sonney2k, will valgrind/backtrace tell me more about that? | 00:25 |
@iglesiasg | no idea what can be wrong, this was working last Friday for sure | 00:26 |
@iglesiasg | arghhhh /* #undef HAVE_EIGEN3 */ | 00:37 |
-!- 17WACBZD1 [~travis-ci@ec2-23-20-31-205.compute-1.amazonaws.com] has joined #shogun | 00:50 | |
17WACBZD1 | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10173358 | 00:50 |
-!- 17WACBZD1 [~travis-ci@ec2-23-20-31-205.compute-1.amazonaws.com] has left #shogun [] | 00:50 | |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has joined #shogun | 00:51 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10174272 | 00:51 |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has left #shogun [] | 00:51 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 01:03 | |
-!- hushell [~hushell@8-92.ptpg.oregonstate.edu] has quit [Ping timeout: 268 seconds] | 01:53 | |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has quit [Ping timeout: 256 seconds] | 02:06 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 02:47 | |
shogun-buildbot | build #423 of nightly_all is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/423 | 03:00 |
shogun-buildbot | build #433 of nightly_none is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/433 | 03:00 |
shogun-buildbot | build #488 of nightly_default is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/488 | 03:12 |
-!- zxtx [~zv@wsip-64-147-1-66.oc.oc.cox.net] has joined #shogun | 03:39 | |
-!- hushell [~hushell@c-24-21-169-136.hsd1.or.comcast.net] has joined #shogun | 03:42 | |
-!- zxtx [~zv@wsip-64-147-1-66.oc.oc.cox.net] has quit [Ping timeout: 246 seconds] | 07:00 | |
-!- gsomix_ [~gsomix@95.67.181.204] has joined #shogun | 07:41 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 07:43 | |
shogun-notifier- | shogun: Viktor Gal :develop * 3e8b5de / CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/3e8b5de49355d2a118b16c89493fc55cf6e96f78 | 07:43 |
shogun-notifier- | shogun: Fix malloc replacement detection and library flags | 07:43 |
shogun-notifier- | shogun: Viktor Gal :develop * 73dd891 / cmake/FindHoard.cmake: https://github.com/shogun-toolbox/shogun/commit/73dd8916a49db134d04da58057c2785ef2be3445 | 07:43 |
shogun-notifier- | shogun: Fix FindHoard.cmake script | 07:43 |
shogun-notifier- | shogun: Viktor Gal :develop * 8754fea / cmake/FindRubyNArray.cmake: https://github.com/shogun-toolbox/shogun/commit/8754fea56a025714ec0100dd88707c93385c930d | 07:43 |
shogun-notifier- | shogun: Fix typo in FindRubyNArray.cmake | 07:43 |
-!- gsomix [~gsomix@80.234.59.144] has quit [Ping timeout: 264 seconds] | 07:43 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 07:44 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 07:44 | |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 07:46 |
shogun-buildbot | build forced [ETA 5m28s] | 07:46 |
shogun-buildbot | I'll give a shout when the build finishes | 07:46 |
shogun-buildbot | build #1503 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1503 | 07:47 |
shogun-buildbot | build #1772 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1772 blamelist: Viktor Gal <viktor.gal@maeth.com> | 07:53 |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has joined #shogun | 08:02 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10183882 | 08:02 |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has left #shogun [] | 08:02 | |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 08:04 |
shogun-buildbot | build forced [ETA 5m28s] | 08:04 |
shogun-buildbot | I'll give a shout when the build finishes | 08:04 |
shogun-buildbot | build #1504 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1504 | 08:04 |
shogun-buildbot | build #1771 of deb1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1771 blamelist: Viktor Gal <viktor.gal@maeth.com> | 08:06 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * f0238e2 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/f0238e23bea93658868e999b4f64599cfcc9cb74 | 08:18 |
shogun-notifier- | shogun: fix build errors / warnings | 08:18 |
-!- lisitsyn1 [~lisitsyn@fb2-lo1.global63.net] has quit [Quit: Leaving.] | 08:58 | |
shogun-buildbot | build #1773 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1773 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:04 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has joined #shogun | 09:06 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10184616 | 09:06 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has left #shogun [] | 09:06 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 09:23 | |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has joined #shogun | 10:35 | |
sonne|work | hmmhh InferenceMethod.get_log_ml_estimate_binary_logit_laplace fails | 10:38 |
-!- gsomix_ [~gsomix@95.67.181.204] has quit [Read error: Connection reset by peer] | 10:48 | |
-!- gsomix__ [~gsomix@95.67.156.101] has joined #shogun | 10:48 | |
-!- gsomix_ [~gsomix@95.67.148.34] has joined #shogun | 10:55 | |
-!- gsomix__ [~gsomix@95.67.156.101] has quit [Ping timeout: 264 seconds] | 10:57 | |
@wiking | sonne|work: mmm what's with java again? | 10:59 |
sonne|work | wiking: no idea but I cannot access the buildbot from work | 10:59 |
sonne|work | wiking: but I added the export JAVA_HOME and IIRC yesterday it worked?! | 10:59 |
@wiking | sonne|work: can i access it or it's a different machine than masteR? | 10:59 |
sonne|work | wiking: you can access it | 11:00 |
@wiking | so it's 7nn right? | 11:00 |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 11:01 |
shogun-buildbot | build forced [ETA 5m28s] | 11:01 |
shogun-buildbot | I'll give a shout when the build finishes | 11:01 |
@wiking | oh shit no | 11:01 |
@wiking | :( | 11:01 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 11:01 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 11:01 | |
shogun-buildbot | build #1505 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1505 | 11:01 |
@wiking | sonne|work: where's the slave stuff? | 11:02 |
@wiking | buildbot's slaves dir is empty | 11:02 |
sonne|work | wiking: I guess buildlsave? | 11:02 |
sonne|work | wiking: ls /home ? | 11:02 |
@wiking | d /home/buildslave/deb3_-_modular_interfaces/ | 11:03 |
@wiking | -bash: cd: /home/buildslave/deb3_-_modular_interfaces/: Permission denied | 11:03 |
@wiking | :( | 11:03 |
@wiking | both with my own account and buildbot accoutn | 11:03 |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 11:03 |
shogun-buildbot | build forced [ETA 5m28s] | 11:03 |
shogun-buildbot | I'll give a shout when the build finishes | 11:03 |
shogun-buildbot | build #1506 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1506 | 11:04 |
@wiking | sonne|work: do u know how to logscale boxplots in d3js? | 11:08 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 11:08 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 11:08 | |
@wiking | iglesiasg: mmm did yesterday's problem got solved with the python script? | 11:08 |
@iglesiasg | wiking, yes. It was related to eigen not being accepted due to its version | 11:09 |
@wiking | iglesiasg: ah ok cool... is 3.1.2 too new? | 11:10 |
@wiking | could we lower the version? | 11:10 |
@iglesiasg | wiking, I am doing now bundle eigen in cmake | 11:10 |
@iglesiasg | so for me it is fine | 11:10 |
@wiking | ok | 11:10 |
-!- foulwall [~user@2001:da8:215:503:7053:1317:a70f:1280] has joined #shogun | 11:14 | |
@wiking | sonne|work: this is really strange | 11:15 |
@wiking | i just cloned and ran: cmake -DRubyModular=ON .. | 11:15 |
@wiking | on 7nn.de | 11:15 |
@wiking | and it found ruby narray | 11:15 |
@wiking | this exact command on 7nn.de ran successfully: cmake -DENABLE_TESTING=ON -DBUNDLE_EIGEN=ON -DBUNDLE_JSON=ON -DPythonModular=ON -DRubyModular=ON -DLuaModular=ON -DRModular=ON -DOctaveModular=ON -DCSharpModular=ON -DJavaModular=ON | 11:17 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 11:18 | |
thoralf | Hmm, I'm getting a "double free or corruption (fasttop): 0x0000000001f02f80" when SG_UNREFing a LibLinear object. | 11:19 |
@wiking | thoralf: HEAD? | 11:19 |
thoralf | It's not reproducible in valgrind/gdb. | 11:19 |
thoralf | wiking: No, friday. Why? | 11:19 |
@wiking | thoralf: then i dont know... :P in head there were some changes with ref/unref so that wouldn't surprise me | 11:20 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 11:28 | |
shogun-notifier- | shogun: Viktor Gal :develop * c348315 / examples/undocumented/python_modular/ (2 files): https://github.com/shogun-toolbox/shogun/commit/c3483151bffbed6d1a60d2e4a04294316a576f73 | 11:28 |
shogun-notifier- | shogun: fix typo in classifier_domainadaptationsvm_modular.py | 11:28 |
shogun-notifier- | shogun: fix imports from numpy in classifier_custom_kernel_modular.py | 11:28 |
thoralf | wiking: No, my HEAD is before these changes. | 11:34 |
shogun-buildbot | build #1774 of deb1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1774 | 11:40 |
shogun-buildbot | build #1388 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1388 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:40 |
shogun-buildbot | build #1205 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1205 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:43 |
shogun-buildbot | build #882 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/882 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:44 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 11:46 | |
shogun-buildbot | build #1418 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1418 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:48 |
@wiking | sonne|work: ok slave4 seems to get it righthttp://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1507 | 11:50 |
@wiking | sonne|work: i mean http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1507 | 11:50 |
sonne|work | wiking: what was it ? | 11:52 |
@wiking | sonne|work: i think slave3 does not have narray... or i dont know | 11:52 |
@wiking | sonne|work: there was a typo in FindRubyNarray.cmake | 11:52 |
@wiking | and that fixed the problem | 11:53 |
@wiking | sonne|work: still the wrong pythonlib will cause some troubles on slave3 | 11:53 |
shogun-buildbot | build #1507 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1507 | 11:53 |
@wiking | oookeeey | 11:54 |
@wiking | well done slave3 | 11:54 |
@wiking | let's define your tests ;) | 11:54 |
thoralf | Is there a SVM in shogun, which can be re-trained? If I get (for example) more data and want to improve an existing SVM? | 11:56 |
sonne|work | wiking: slave 3 is an older debian, slave 4 a current one | 12:03 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 12:04 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 12:04 | |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 12:04 |
shogun-buildbot | build forced [ETA 31m57s] | 12:04 |
shogun-buildbot | I'll give a shout when the build finishes | 12:04 |
@wiking | sonne|work: http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1508 | 12:14 |
-!- travis-ci [~travis-ci@ec2-23-20-31-205.compute-1.amazonaws.com] has joined #shogun | 12:22 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10189972 | 12:22 |
-!- travis-ci [~travis-ci@ec2-23-20-31-205.compute-1.amazonaws.com] has left #shogun [] | 12:22 | |
-!- foulwall [~user@2001:da8:215:503:7053:1317:a70f:1280] has quit [Ping timeout: 245 seconds] | 12:34 | |
shogun-buildbot | build #1508 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1508 | 12:38 |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has quit [Quit: Leaving.] | 12:42 | |
shogun-notifier- | shogun: Viktor Gal :develop * 8083aa8 / / (2 files): https://github.com/shogun-toolbox/shogun/commit/8083aa84d573fb375d52ec34284f3d2be6c57020 | 12:45 |
shogun-notifier- | shogun: Attempt to fix issue #1408 | 12:45 |
shogun-notifier- | shogun: Remove import from numpy rand and seed in classifier_custom_kernel_modular.py | 12:45 |
@wiking | shogun-buildbot: force build --branch=develop 'cyg1 - libshogun' | 12:48 |
shogun-buildbot | build forced [ETA 16m21s] | 12:48 |
shogun-buildbot | I'll give a shout when the build finishes | 12:48 |
-!- foulwall [~user@2001:da8:215:6901:c0cd:22cc:a3d0:e7b4] has joined #shogun | 12:51 | |
shogun-buildbot | build #1206 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1206 | 12:54 |
@wiking | oh that internet is not really the most reliable or something is wrong with md5 | 12:56 |
thoralf | wiking: Does cmake check for new google test/mock versions every time I build? | 13:05 |
thoralf | wiking: This would break my offline workflow... ;) | 13:06 |
@wiking | ooooh shiiiiiiiiiiiit | 13:06 |
@wiking | i have a problem | 13:07 |
@wiking | that 1 megabyte eigen release costed me about 11TB of traffic this month! | 13:07 |
@wiking | i need to move that somewhere because my monthly traffic limit is 20TB | 13:08 |
-!- foulwall` [~user@2001:da8:215:503:61a7:220a:ae70:ee95] has joined #shogun | 13:09 | |
thoralf | wiking: 11TB? How could that happen? | 13:10 |
lisitsyn | TB? | 13:10 |
thoralf | Would be 11.000.000 downloads. | 13:11 |
@wiking | lisitsyn: yes | 13:11 |
-!- foulwall [~user@2001:da8:215:6901:c0cd:22cc:a3d0:e7b4] has quit [Ping timeout: 264 seconds] | 13:11 | |
@wiking | 11,514.525 gigabytes | 13:12 |
sonne|work | wiking: how that? | 13:12 |
@wiking | to be precise | 13:12 |
@wiking | well | 13:12 |
@wiking | travis? :) | 13:12 |
@wiking | each time there's a commit or PR | 13:12 |
@wiking | it downloads the thing like 10 times | 13:12 |
sonne|work | wiking: well 1000 builds -> 1 GB! | 13:12 |
@wiking | sorry 11 times | 13:12 |
@wiking | as there are 11 jobs in one build on travis | 13:12 |
@wiking | so | 13:13 |
@wiking | we need to find a solution for this | 13:13 |
@wiking | nowish :) | 13:13 |
lisitsyn | crazy! | 13:13 |
thoralf | wiking: As I written above: This downloading/updating breaks my offline/mobile workflow, too.# | 13:14 |
@wiking | thoralf: cannot do anything about that | 13:14 |
@wiking | thoralf: ask google ppl to do a release | 13:15 |
@wiking | or | 13:15 |
sonne|work | wiking: well we could for sure - just add an option to specify the path to google mock/test | 13:15 |
@wiking | sonne|work: i think we've discussed this | 13:15 |
@wiking | sonne|work: there's no way we can check the version of that gmock | 13:15 |
sonne|work | wiking: except for a compile test ... | 13:16 |
@wiking | sonne|work: i'm happy to see patches for this | 13:16 |
-!- foulwall` [~user@2001:da8:215:503:61a7:220a:ae70:ee95] has quit [Remote host closed the connection] | 13:16 | |
lisitsyn | we can add a submodule | 13:17 |
lisitsyn | in git | 13:17 |
@wiking | lisitsyn: svn submodule? :) | 13:17 |
lisitsyn | wiking: well we can set up some git stuff for that | 13:17 |
@wiking | yeah | 13:18 |
@wiking | as said earlier happy to see patches | 13:18 |
sonne|work | lisitsyn: excellent idea! | 13:18 |
@wiking | as well we need a mirror for eigen | 13:19 |
sonne|work | lisitsyn: much better than pulling external stuff all the time with svn and from other sources | 13:19 |
sonne|work | wiking: no we could just have a shogun-external git submodule | 13:19 |
sonne|work | that could contain all the external dependencies | 13:19 |
@wiking | sonne|work: i'm good with that as well | 13:19 |
sonne|work | that then cmake could use (git pull ...) | 13:19 |
@wiking | just need the solution today | 13:19 |
thoralf | Yeah, perfect to control versions of dependencies. | 13:20 |
sonne|work | thoralf: yes exactly | 13:20 |
@wiking | thoralf: it's already controlled this way | 13:20 |
thoralf | wiking: No exactly - since it updates automatically, it could break on upstream updates, right? | 13:20 |
thoralf | upstream as-in google test updates. | 13:20 |
@wiking | thoralf: no | 13:21 |
@wiking | thoralf: because it's fixed to a revision | 13:21 |
thoralf | wiking: I see. | 13:21 |
@wiking | i doesn't just fetches HEAD | 13:21 |
@wiking | sonne|work: did we run integration tests on deb3 - modular_interfaces | 13:23 |
@wiking | ? | 13:23 |
sonne|work | wiking: yes sure | 13:24 |
@wiking | ok | 13:24 |
sonne|work | wiking: how difficult is it to do a git clone shogun-external and then use the extracted sources (or .tar.gz) in there if a dependency could not be found? | 13:25 |
sonne|work | wiking: with cmake I mean | 13:25 |
thoralf | wiking: If you fixed the revision of google test/mock, why are you updating it on every build? | 13:25 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 13:26 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 13:26 | |
@wiking | thoralf: don't ask me ask cmake ppl | 13:26 |
@wiking | thoralf: maybe adding UPDATE_COMMAND "" would help actually | 13:28 |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has joined #shogun | 13:28 | |
travis-ci | [travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10192551 | 13:28 |
-!- travis-ci [~travis-ci@ec2-67-202-44-165.compute-1.amazonaws.com] has left #shogun [] | 13:28 | |
@wiking | sonne|work: why would we do that? | 13:28 |
@wiking | sonne|work: i mean if it's a submodule then we should just assume that there are the sources | 13:29 |
@wiking | or tar.gz/bz2 and that's it | 13:29 |
sonne|work | wiking: because we normally want to use the system's libs | 13:32 |
@wiking | sonne|work: yes my problem wasn't that | 13:32 |
@wiking | sonne|work: it was about why would we want to do a git clone? | 13:32 |
@wiking | sonne|work: if dependency is not met | 13:32 |
@wiking | it should be there | 13:33 |
sonne|work | wiking: to fetch the dependencies | 13:33 |
thoralf | wiking: I didn't find a place to add UPDATE_COMMAND. | 13:33 |
sonne|work | wiking: ahh you mean you would manually require people to issue git submodule update shogun-external ? | 13:33 |
@wiking | sonne|work: and why not do that with git submodule init/update | 13:33 |
@wiking | sonne|work: yes | 13:33 |
sonne|work | sure also possible | 13:34 |
@wiking | thoralf: nevermind... just testing | 13:34 |
@wiking | ok this works it seem | 13:34 |
@wiking | *seems | 13:34 |
@wiking | so i have a solution for not always updating svn | 13:34 |
@wiking | just the one time checkout | 13:34 |
@wiking | can we live with that? | 13:34 |
thoralf | wiking: I could. | 13:35 |
@wiking | but still | 13:35 |
@wiking | i need a non https mirror for eigen | 13:35 |
shogun-buildbot_ | build #1775 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1775 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:36 |
sonne|work | wiking: why that? | 13:36 |
sonne|work | wiking: lets just switch to doing that | 13:36 |
@wiking | sonne|work: cmake cannot handle https by default | 13:37 |
sonne|work | wiking: no I mean we use the submodule idea now | 13:37 |
@wiking | sonne|work: but now it's all fixed... | 13:37 |
@wiking | only the mirror is required | 13:37 |
@wiking | almost | 13:40 |
@wiking | i have found eigen-3.1.3 mirror among gentoo's distfiles | 13:42 |
@wiking | ok | 13:43 |
@wiking | http://ubuntu.osuosl.org/ubuntu/pool/universe/e/eigen3/eigen3_3.1.2.orig.tar.bz2 | 13:43 |
@wiking | can we live with this? | 13:43 |
@wiking | i mean now we have 3.1.4 | 13:43 |
@wiking | but we require minimum 3.1.2 | 13:43 |
@wiking | so actually having 3.1.2 provided should be good enough | 13:43 |
@wiking | i'm changing | 13:43 |
@wiking | as my bandwidth is burning | 13:43 |
sonne|work | wiking: or that http://ftp.de.debian.org/debian/pool/main/e/eigen3/eigen3_3.2.0.orig.tar.bz2 | 13:45 |
-!- thoralf [~thoralf@enki.zib.de] has quit [Ping timeout: 248 seconds] | 13:45 | |
@wiking | okey pushed | 13:47 |
shogun-notifier- | shogun: Viktor Gal :develop * bd16463 / src/interfaces/r_modular/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/bd16463f6474a2295243172b0630174b928ac5d9 | 13:47 |
shogun-notifier- | shogun: Fix compiler flags for RModular | 13:47 |
shogun-notifier- | shogun: Viktor Gal :develop * 0762437 / cmake/external/Eigen3.cmake,cmake/external/GoogleTestNMock.cmake: https://github.com/shogun-toolbox/shogun/commit/076243773888326ed878027b666d85794cbcaab8 | 13:47 |
shogun-notifier- | shogun: Disable update command of GMOCK for not updating always the checked out | 13:47 |
shogun-notifier- | shogun: svn repository | 13:47 |
shogun-notifier- | shogun: Switch to another eigen mirror as it's currently burning my server's | 13:47 |
shogun-notifier- | shogun: bandwidth | 13:47 |
@wiking | now we can really know who of us uses latest develop branch of shogun (/me is looking at the apache access logs :P) | 13:48 |
@wiking | i'll give it a day more before removing the archive from there... | 13:49 |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 13:50 |
@wiking | shogun-buildbot_: force build --branch=develop 'deb3 - modular_interfaces' | 13:51 |
shogun-buildbot_ | The build has been queued, I'll give a shout when it starts | 13:51 |
-!- thoralf [~thoralf@enki.zib.de] has joined #shogun | 13:52 | |
shogun-buildbot_ | build #1776 of deb1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1776 | 13:53 |
shogun-buildbot_ | build forced [ETA 41m36s] | 13:53 |
shogun-buildbot_ | I'll give a shout when the build finishes | 13:53 |
shogun-buildbot_ | build #883 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/883 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:55 |
@wiking | sonne|work: so once the modular is done | 13:58 |
@wiking | sonne|work: can i set up the nightly to generate tar.bz2 and tar.gz packages? | 13:58 |
@wiking | sonne|work: if so, where can we upload the generated files? | 13:58 |
shogun-buildbot_ | build #1419 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1419 blamelist: Viktor Gal <viktor.gal@maeth.com> | 13:59 |
sonne|work | wiking: just build them and then I can create a shell script to put them somewhere | 13:59 |
@wiking | sonne|work: and what's the news with that osx buildbot? | 13:59 |
@wiking | sonne|work: yeah i thought to help u out and write u the shellscript | 14:00 |
sonne|work | well we currently have other issues so I didn't have time to look into that | 14:00 |
@wiking | okey... | 14:00 |
shogun-buildbot_ | build #1207 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1207 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:00 |
@wiking | sonne|work: http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1207/steps/compile/logs/stdio | 14:00 |
@wiking | sonne|work: dontunderstand... | 14:00 |
sonne|work | no the shell script is just few mins work for me (not the cmake though :/) | 14:00 |
@wiking | sonne|work: the cmake for the packages are actually only: make package | 14:00 |
@wiking | and it'll generate the files under ./build | 14:01 |
sonne|work | wiking: I guess it has no permissions to download anything with whatever too you used | 14:01 |
thoralf | actual MD5 sum: [d41d8cd98f00b204e9800998ecf8427e] | 14:06 |
thoralf | this is the md5 of en empty file :) | 14:06 |
sonne|work | s/too/tool | 14:10 |
@wiking | sonne|work: it's integrated in cmake | 14:11 |
@wiking | but yeah maybe network activity is disabled on windows or something | 14:11 |
sonne|work | wiking: yeah then cmake is blocked by some firewall | 14:11 |
shogun-notifier- | shogun: Roman Votyakov :develop * 71737a1 / tests/unit/machine/gp/InferenceMethod_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/71737a122830391eca3a7ff1de35d979e4e8104e | 14:15 |
shogun-notifier- | shogun: decrease accuracy of get_log_ml_estimate_binary_logit_laplace test for now | 14:15 |
shogun-notifier- | shogun: Heiko Strathmann :develop * cb022d0 / tests/unit/machine/gp/InferenceMethod_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/cb022d018134d550405a25cc375af0935eb129d5 | 14:15 |
shogun-notifier- | shogun: Merge pull request #1410 from votjakovr/feature/gp_refactoring | 14:15 |
shogun-notifier- | shogun: | 14:15 |
shogun-notifier- | shogun: Decrease accuracy of get_log_ml_estimate_binary_logit_laplace test for now | 14:15 |
-!- iglesiasg [~iglesias@2001:6b0:1:1da0:3dd0:6deb:381f:6e88] has joined #shogun | 14:37 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 14:37 | |
shogun-buildbot_ | build #1389 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1389 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:38 |
shogun-buildbot_ | build #1510 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1510 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:39 |
shogun-buildbot_ | build #884 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/884 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:40 |
shogun-buildbot_ | build #1509 of deb3 - modular_interfaces is complete: Failure [failed install test python_modular test lua_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1509 | 14:40 |
@wiking | sonne|work: heh actually modular interface bot is almost done \o/ | 14:41 |
shogun-buildbot_ | build #1390 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1390 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:41 |
shogun-buildbot_ | build #1208 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1208 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:44 |
shogun-buildbot_ | build #1420 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1420 blamelist: Viktor Gal <viktor.gal@maeth.com> | 14:44 |
@wiking | sonne|work: http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1509/steps/install/logs/stdio | 14:54 |
@wiking | sonne|work: check the end of the log... where should we install the ruby modular? | 14:54 |
sonne|work | wiking: I don't remember best is to check configure or some binary ruby package | 15:14 |
@wiking | sonne|work: it should be there | 15:18 |
@wiking | sonne|work: the problem is the permission | 15:18 |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has joined #shogun | 15:24 | |
shogun-buildbot_ | build #1391 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1391 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:25 |
shogun-buildbot_ | build #885 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/885 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:27 |
shogun-buildbot_ | build #1421 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1421 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:30 |
shogun-buildbot_ | build #1209 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1209 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:32 |
thoralf | sonne|work, wiking: I think CRandom is not thread safe. | 15:34 |
thoralf | We have one global instance of CRandom, but different threads may be using it. | 15:34 |
@wiking | sonne|work: it is | 15:35 |
@wiking | thoralf: sorry it is | 15:35 |
@wiking | thoralf: but then again i might overlooked something so plz show me where does it go wrong | 15:35 |
thoralf | wiking: Okay, do we agree that there is only one global instance of CRandom and that - if using them from threads - all threads are accessing the same instance? | 15:39 |
@wiking | thoralf: maybe adding an openmp based unittest for CRandom would be a good idea... i'll be waiting for the PR | 15:39 |
@wiking | thoralf: yes | 15:39 |
sonne|work | thoralf: yes that function is *not* intended to be thread safe. if you want thread safe stuff use a Random object | 15:40 |
sonne|work | per thread | 15:40 |
thoralf | sonne|work: Okay, thank you. | 15:40 |
thoralf | sonne|work: I'm learning different machines in parallel using threads - and each thread uses stratifiedcrossvalidation, which uses the global crandom state. | 15:41 |
thoralf | sonne|work: Do you see how to avoid this? | 15:42 |
sonne|work | thoralf: only way would be to change stratifiedcrossvalidation to take a CRandom object as input | 15:42 |
thoralf | I'm getting strange errors and I think I have tracked it down to "shogun::CRandom::random_32() const (Random.cpp:123)" | 15:42 |
thoralf | Okay, I'll try that. | 15:43 |
sonne|work | thoralf: maybe if that object is null take the default one from CMath or so | 15:44 |
thoralf | Won't be easy, because StratifiedCrossValidationSplitting.cpp uses current->shuffle(), so I have to pass RNG states around. | 15:46 |
thoralf | CDynamicArray::shuffle calls CDynArray::shuffle calls CMath::random calls sg_rand->random() | 15:50 |
thoralf | Jesus. ;) | 15:50 |
@wiking | thoralf sonne|work imho we should start deprecating CMath::random* stuff | 15:51 |
@wiking | it's just an unecessary extra wrapper... | 15:51 |
thoralf | wiking: How does it solve the problem? | 15:51 |
@wiking | thoralf: your's it doesn't solve | 15:51 |
@wiking | thoralf: i've just remembered now when u wrote this... | 15:52 |
thoralf | wiking: Yeah, it's quite nested. | 15:52 |
sonne|work | wiking: it is convenient though to just generate random numbers without having to create an object | 15:52 |
@wiking | thoralf: u can make the CRandom thread safe quite easily.... as there's already a mutex in that class | 15:52 |
sonne|work | wiking: like java's Random and Math.random | 15:52 |
@wiking | sonne|work: yeah sg_rand is there | 15:52 |
sonne|work | one has lots of features the other is just the simple thing | 15:52 |
@wiking | sonne|work: no need to create an object as it's there as global... | 15:53 |
thoralf | wiking: But doesn't a mutex syncronize my threads and will kill parallelism? | 15:53 |
@wiking | thoralf: well it's just locks until it gives u back the random... | 15:53 |
@wiking | thoralf: i mean there's simply no other means to assure thread safetiness | 15:54 |
thoralf | wiking: Yeah, but I'm needing loooots of random numbers. | 15:54 |
@wiking | thoralf: but the locks will be quite short -> it's not going to be unparallelized ;) | 15:54 |
thoralf | There is: One state for each thread. | 15:54 |
@wiking | thoralf: in that case as sonne|work said you are better of with a CRandom for each thread | 15:55 |
@wiking | sonne|work: do we have a common way in shogun to mark deprecation? | 15:55 |
sonne|work | wiking: no but I don't think we should deprecate CMath::random | 15:56 |
@wiking | i see SGIO.h | 15:56 |
@wiking | sonne|work: why? | 15:56 |
thoralf | Yes, that's what I've been thinking. But I needed to make sure that I'm not completly wrong. :) | 15:56 |
thoralf | wiking | 15:56 |
sonne|work | wiking: because it is easy to generate a random number with that w/o creating an object | 15:56 |
shogun-buildbot_ | build #1511 of deb3 - modular_interfaces is complete: Failure [failed install test python_modular test lua_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1511 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 15:57 |
@wiking | sonne|work: but one doesn't need to create an object for random w/o CMath::random | 15:58 |
@wiking | sonne|work: as sg_rand i there by default | 15:58 |
@wiking | *is there | 15:59 |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 15:59 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 15:59 | |
@wiking | ok at least this function should be deprecated or changed: CMath::normal_random | 16:00 |
sonne|work | wiking: yes sure that would work too but I prefer CMath::random | 16:00 |
@wiking | as there's a faster implementation of it in CRandom | 16:00 |
sonne|work | wiking: better drop it | 16:00 |
sonne|work | wiking: do we have a test for that (I mean that this is really gaussian?) | 16:01 |
@wiking | sonne|work: for CMath::random no | 16:01 |
@wiking | i mean for CMath::normal_random no | 16:01 |
sonne|work | IIRC I wrote sth fro uniform | 16:01 |
sonne|work | dont' remember for gaussian | 16:01 |
@wiking | for CRandom::normal_distrib yes | 16:02 |
@wiking | sonne|work: Random::random_std_normal_quantiles unittest | 16:02 |
@wiking | that actually tests the CRandom's normal_distrib | 16:02 |
sonne|work | alright then die normal_random! | 16:03 |
@wiking | mmm .cpuinfo.c should die too | 16:10 |
van51 | why isn't SGReferencedData copy constructor calling copy_data() ? | 16:17 |
van51 | like overloaded operator= does | 16:18 |
@sonney2k | van51, you cannot call a virtual function from the constructor | 16:22 |
@sonney2k | wiking, I don't have that file | 16:23 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 58723bc / .clang_complete,src/.clang_complete: https://github.com/shogun-toolbox/shogun/commit/58723bcb3990d583d4c44d8c4c00afd7b2b322bc | 16:25 |
shogun-notifier- | shogun: move clang complete to new root dir | 16:25 |
@sonney2k | wiking, why is it complaining about python_modular in the lua_modular interface? http://shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1511/steps/test%20lua_modular/logs/stdio | 16:25 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 16:29 | |
@sonney2k | shogun-buildbot, force build --branch=develop 'deb3 - modular_interfaces' | 16:31 |
shogun-buildbot | build forced [ETA 41m36s] | 16:31 |
shogun-buildbot | I'll give a shout when the build finishes | 16:31 |
@sonney2k | iglesiasg, wiking, lisitsyn - please add third party dependencies in shogun-external ( a new repo https://github.com/shogun-toolbox/shogun-external) | 16:36 |
@sonney2k | I suggest to use name/release.tgz | 16:37 |
@iglesiasg | sonney2k, we will use it as a submodule from the main project, right? | 16:37 |
@sonney2k | so we can remove automatic download stuff from cmake and have a consistently tested environment | 16:37 |
@sonney2k | iglesiasg, yes we will create an external dir and add it as submodule | 16:38 |
@iglesiasg | sonney2k, I am not sure if I like better cmake taking care of that | 16:38 |
@iglesiasg | the con I see with it is that AFAIK submodules have to be added by hand | 16:39 |
@iglesiasg | sure, it is just git submodule init && git submodule update | 16:39 |
@sonney2k | iglesiasg, the problem is that we dont' want to all the time fetch the data | 16:39 |
@sonney2k | s/data/externals/ | 16:39 |
@sonney2k | and work offline | 16:39 |
@iglesiasg | sonney2k, yeah, that is a bummer | 16:39 |
@sonney2k | impossible currently | 16:39 |
@sonney2k | gtg | 16:40 |
@sonney2k | cu | 16:40 |
@iglesiasg | I don't even understand why it has to download each time | 16:40 |
@iglesiasg | see you! | 16:40 |
thoralf | Hes iglesiasg | 16:48 |
@iglesiasg | hi thoralf! | 16:48 |
-!- FSCV [~FSCV@50.7.50.60] has joined #shogun | 16:50 | |
van51 | how can I run just a specific unit test now? | 17:02 |
pickle27 | van51: same as before but the exe is in /tests under your build dir | 17:03 |
van51 | pickle27: ah indeed | 17:04 |
van51 | pickle27: thank you :) | 17:04 |
pickle27 | van51: np | 17:04 |
shogun-buildbot | build #1512 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1512 | 17:04 |
shogun-buildbot | build #1392 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1392 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:16 |
shogun-buildbot | build #886 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/886 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:18 |
shogun-buildbot | build #1422 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1422 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:21 |
shogun-buildbot | build #1210 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1210 blamelist: Soeren Sonnenburg <sonne@debian.org> | 17:23 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has joined #shogun | 17:25 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10200283 | 17:25 |
-!- travis-ci [~travis-ci@ec2-54-225-61-153.compute-1.amazonaws.com] has left #shogun [] | 17:25 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 17:43 | |
-!- gsomix_ [~gsomix@95.67.148.34] has quit [Ping timeout: 245 seconds] | 17:50 | |
-!- gsomix_ [~gsomix@95.67.139.30] has joined #shogun | 17:55 | |
-!- gsomix_ [~gsomix@95.67.139.30] has quit [Ping timeout: 268 seconds] | 18:00 | |
@wiking | sonney2k: i dislike this idea | 18:01 |
@wiking | sonney2k: and now the fetching is fixed | 18:01 |
@wiking | sonney2k: namely everything is once fetched to unders <shogun root>/third_parties | 18:01 |
@wiking | it is essentially the same thing as having another git repo | 18:01 |
@wiking | but this one circuments any emails | 18:02 |
@wiking | that people forget to init submodules | 18:02 |
@wiking | as they usually do (see the emails about not having data for the examples...) | 18:02 |
@wiking | it is not impossible it has been fixed... | 18:02 |
@iglesiasg | wiking, that was exactly the idea why I don't like to add submodules even to compile | 18:12 |
@wiking | ok i need to make one more little change that this is going to be actually true | 18:13 |
@wiking | as currently the GoogleMock src is populated under the build dir | 18:13 |
@wiking | so if you rm -rf * the build | 18:14 |
@wiking | then it'll have to checkout again the src | 18:14 |
@wiking | i mean the svn | 18:14 |
@wiking | but this is now the only remaining thing | 18:14 |
@wiking | to have a totally offline cmake based build system | 18:14 |
-!- gsomix_ [~gsomix@95.67.139.30] has joined #shogun | 18:17 | |
@wiking | email sent. | 18:26 |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 19:03 | |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has quit [Quit: Leaving.] | 19:12 | |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has joined #shogun | 19:13 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 19:25 | |
@sonney2k | wiking, iglesiasg - if a dependency is missing you can easily do a git submodule update to fetch it | 19:40 |
@sonney2k | so no emails to mailinglists | 19:40 |
@sonney2k | we don't even need it as submodule | 19:40 |
@iglesiasg | I think that we will get some mails like the ones we get about shogun-data | 19:41 |
@iglesiasg | I know it is easy to do git submodule update :) | 19:41 |
@sonney2k | iglesiasg, no nothing manual in there | 19:42 |
@sonney2k | iglesiasg, if some dependency is missign we just fetch them from inside cmake | 19:43 |
@sonney2k | but you don't have to have subversion installed to do so | 19:43 |
@wiking | sonney2k: anyways this is really for testing | 19:53 |
@wiking | sonney2k: i cannot see the use case when a developer who wants to run unit test does not have svn | 19:54 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 19:54 | |
shogun-notifier- | shogun: Thoralf Klein :develop * 9d8c3f6 / src/shogun/lib/SGSparseVector.cpp: https://github.com/shogun-toolbox/shogun/commit/9d8c3f681994bce568e458489a65a94a3c768219 | 19:54 |
shogun-notifier- | shogun: * fixed invalid read in SGSparseVector::sort_features(): Worked on realloc'ed memory. | 19:54 |
shogun-notifier- | shogun: * SGSparseVector::sort_features(): freeing allocated memory as soon as possible | 19:54 |
shogun-notifier- | shogun: Thoralf Klein :develop * 68e9b70 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/68e9b704546a54e1abe509c6b351cf5f2ca486d6 | 19:54 |
shogun-notifier- | shogun: * Introduced clone() method for SGSparseVector (similar to SGVector). Added unit test. | 19:54 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 1479f95 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/1479f95e8f55865c4d10313e3c47f087a86dafed | 19:54 |
shogun-notifier- | shogun: Merge pull request #1412 from tklein23/sparse_fixes | 19:54 |
shogun-notifier- | shogun: | 19:54 |
shogun-notifier- | shogun: SGSparseVector: fixed invalid reads in sort_features() and introduce clone() method + unit test | 19:54 |
@wiking | let along a distrib builder | 19:54 |
@wiking | who builds packages for distribs | 19:54 |
@sonney2k | wiking, well we need a way to do offline builds like we had | 19:55 |
@sonney2k | van51, I am currently not sure how the SGReferencedData code was doing it before | 19:56 |
@sonney2k | van51, IIRC we had a pointer to a *single* refcount object for a vector (and its clones) | 19:56 |
@wiking | sonney2k: the thing i've just described in the email does allow u that | 19:57 |
@wiking | sonney2k: essentially it is down to one thing: you get the externals with git or svn | 19:57 |
@wiking | and basically now this discussion became about what is better: git or svn | 19:57 |
@sonney2k | van51, since a SGReferenced* object is only destroyed when refcount == 0 there can be just this single object left | 19:57 |
@sonney2k | van51, and so this one can safely delete the ptr when count == 0 is reached and the vector data | 19:58 |
@wiking | sonney2k: sorry: git vs svn+wget | 19:58 |
shogun-buildbot | build #1393 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1393 blamelist: Thoralf Klein <thoralf.klein@zib.de> | 19:58 |
shogun-buildbot | build #1423 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1423 blamelist: Thoralf Klein <thoralf.klein@zib.de> | 19:58 |
@sonney2k | wiking, well no not git svn wget whatever | 19:58 |
@sonney2k | wiking, but have them in one reliable place - git repo - or somewhere distributed | 19:58 |
van51 | sonney2k: yea it's a single object now as well | 19:59 |
-!- iglesiasg [~iglesias@2001:6b0:1:1da0:3dd0:6deb:381f:6e88] has quit [Ping timeout: 245 seconds] | 19:59 | |
van51 | sonney2k: even when the object was copied it maintained that single object right | 20:00 |
van51 | ? | 20:00 |
@sonney2k | wiking, besides it is much easier to fetch this git - no need to adjust download urls etc and we can ensure versions that work well together | 20:00 |
@sonney2k | van51, yes a single refcount object only | 20:00 |
shogun-buildbot | build #887 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/887 blamelist: Thoralf Klein <thoralf.klein@zib.de> | 20:00 |
van51 | sonney2k: yeah that's what I had changed, but I reverted it in the latest PR | 20:01 |
@wiking | sonney2k: that is being assured by a rev... we do have to write for each external project (regardless of where the src coming from) a custom .cmake | 20:01 |
@wiking | that we cannot avoid | 20:01 |
van51 | sonney2k: but should copied SGObjects keep a single refcount object as well? | 20:01 |
@wiking | sonney2k: if we take the submodule way to bump up a version of a given package: 1) we have to update the git submodule 2) we have to update the base shogun git repo. just like data update | 20:03 |
@wiking | instead of one update on the main shogun repo | 20:03 |
shogun-buildbot | build #1211 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1211 blamelist: Thoralf Klein <thoralf.klein@zib.de> | 20:04 |
@sonney2k | van51, what is different with SGObject is that we only pass around pointers | 20:06 |
@sonney2k | van51, so we only need to delete the last object and all is gone | 20:08 |
@sonney2k | van51, I don't think the copy constructors really ever worked... | 20:08 |
@sonney2k | wiking, well we don't *have to* use a submodule but I think it is a good idea to use it since this way we can ensure that a certain version of shogun requires certain set of third party deps | 20:09 |
van51 | sonney2k: the way I have it in the copy constructor now of SGObject is that the new object will share the same ref counter with the original one | 20:11 |
van51 | sonney2k: apart from that I believe that everything else has switched ok | 20:11 |
-!- iglesiasg [~iglesias@2001:6b0:1:1041:916f:6a22:749d:c7e2] has joined #shogun | 20:12 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 20:12 | |
van51 | sonney2k: also we have to decide how to unref | 20:12 |
van51 | sonney2k: bc like you said there could be a race condition if we tested ref_count <= 0 | 20:13 |
van51 | sonney2k: and the way it is now, if you take a look at the unit test case, it requires an additional SG_REF | 20:13 |
@sonney2k | van51, yes but I realized that this cannot be a problem because - when we do SG_UNREF more often then SG_REF it will crash anyways | 20:16 |
@sonney2k | van51, so it is totally ok to check for ref_count <= 0 | 20:16 |
@sonney2k | and then delete this | 20:16 |
@sonney2k | van51, same with SGReferencedData: if we have 2 threads assigning a=b and a=c this is a race condition anyway, which should be solved at some other place | 20:18 |
van51 | sonney2k: ok I get it | 20:19 |
van51 | sonney2k: so I switch back to <= | 20:19 |
van51 | sonney2k: also should I keep that test in the unit test file? it doesn't seem like it fits | 20:20 |
@sonney2k | van51, sry which test? | 20:20 |
@sonney2k | van51, so for SGObject also the same Refcount object right! | 20:21 |
van51 | sonney2k: in the PR, i had added a unit test case to make sure it was working | 20:21 |
@sonney2k | van51, which PR | 20:22 |
van51 | sonney2k: https://github.com/shogun-toolbox/shogun/pull/1406 | 20:22 |
shogun-buildbot | build #1514 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1514 blamelist: Thoralf Klein <thoralf.klein@zib.de> | 20:23 |
-!- iglesiasg [~iglesias@2001:6b0:1:1041:916f:6a22:749d:c7e2] has quit [Quit: Ex-Chat] | 20:28 | |
van51 | sorry but I g2g. | 20:29 |
van51 | I update to using <= 0 | 20:29 |
van51 | hope you and travis approve :) | 20:29 |
van51 | byee | 20:29 |
-!- van51 [~van51@athedsl-224969.home.otenet.gr] has quit [Quit: Leaving.] | 20:29 | |
shogun-buildbot | build #1515 of deb3 - modular_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1515 blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de> | 20:40 |
shogun-buildbot | build #1394 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1394 blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de> | 20:40 |
shogun-buildbot | build #1212 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1212 blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de> | 20:46 |
shogun-buildbot | build #1424 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1424 blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de> | 20:47 |
-!- iglesiasg [~iglesias@2001:6b0:1:1041:7484:554c:fb95:9550] has joined #shogun | 20:48 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 20:48 | |
shogun-buildbot | build #888 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/888 blamelist: Soeren Sonnenburg <sonne@debian.org>, Thoralf Klein <thoralf.klein@zib.de> | 20:50 |
-!- travis-ci [~travis-ci@ec2-107-20-109-156.compute-1.amazonaws.com] has joined #shogun | 20:57 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/10208628 | 20:57 |
-!- travis-ci [~travis-ci@ec2-107-20-109-156.compute-1.amazonaws.com] has left #shogun [] | 20:57 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 21:22 | |
-!- iglesiasg [~iglesias@2001:6b0:1:1041:7484:554c:fb95:9550] has quit [Ping timeout: 245 seconds] | 21:46 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 264 seconds] | 22:36 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 22:54 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 22:57 | |
shogun-notifier- | shogun: van51 :develop * 9dd10a2 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/9dd10a2b060f0228db189af664ff51bfbd16cc6b | 22:57 |
shogun-notifier- | shogun: Made refcounting in SGObject use the RefCount class | 22:57 |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * c2a6ab3 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/c2a6ab3ca9261ffa52f709003f8afac2e85b6abe | 22:57 |
shogun-notifier- | shogun: Merge pull request #1406 from van51/feature/refcount | 22:57 |
shogun-notifier- | shogun: | 22:57 |
shogun-notifier- | shogun: Made refcounting in SGObject use the RefCount class | 22:57 |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has quit [Quit: Leaving.] | 23:05 | |
-!- FSCV [~FSCV@50.7.50.60] has quit [Read error: Connection reset by peer] | 23:08 | |
shogun-buildbot | build #1395 of deb2 - static_interfaces is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb2%20-%20static_interfaces/builds/1395 blamelist: van51 <vangelis_51@hotmail.com> | 23:13 |
shogun-buildbot | build #889 of rpm1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/rpm1%20-%20libshogun/builds/889 blamelist: van51 <vangelis_51@hotmail.com> | 23:13 |
shogun-buildbot | build #1213 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1213 blamelist: van51 <vangelis_51@hotmail.com> | 23:14 |
shogun-buildbot | build #1425 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/bsd1%20-%20libshogun/builds/1425 blamelist: van51 <vangelis_51@hotmail.com> | 23:19 |
shogun-buildbot | build #1214 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1214 | 23:32 |
shogun-buildbot | build #1215 of cyg1 - libshogun is complete: Failure [failed compile] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/1215 | 23:39 |
shogun-buildbot | build #1783 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1783 blamelist: Soeren Sonnenburg <sonne@debian.org> | 23:45 |
shogun-buildbot | build #1516 of deb3 - modular_interfaces is complete: Failure [failed compile r_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1516 blamelist: van51 <vangelis_51@hotmail.com> | 23:55 |
--- Log closed Thu Aug 15 00:00:15 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!