--- Log opened Tue Aug 30 00:00:37 2011 | ||
-!- blackburn1 [~blackburn@188.168.4.53] has quit [Ping timeout: 240 seconds] | 00:39 | |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 03:52 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Quit: Leaving] | 05:05 | |
@sonney2k | serialhex, - you patch is not mergable | 06:16 |
---|---|---|
@sonney2k | you have to rebase it | 06:16 |
-!- blackburn [~blackburn@188.168.3.121] has joined #shogun | 10:12 | |
-!- heiko [~heiko@134.91.52.222] has joined #shogun | 10:20 | |
@sonney2k | morning heiko - glad to see that your gf let you live and code ;-) | 10:21 |
heiko | morning :) | 10:21 |
heiko | yes, me too ;) | 10:21 |
heiko | ok then, now for some hardcore coding action :) | 10:22 |
-!- blackburn [~blackburn@188.168.3.121] has quit [Ping timeout: 260 seconds] | 10:24 | |
-!- blackburn [~blackburn@188.168.3.98] has joined #shogun | 10:25 | |
@sonney2k | heiko, just ping me when there is anything | 10:51 |
heiko | sonney2k, ok i will | 10:51 |
heiko | now it the migrate method has to return by reference whether a new TParameter instance is returned, and also if its was was newly allocated | 10:52 |
heiko | so all cases may be handleed | 10:52 |
@sonney2k | ok | 10:56 |
@sonney2k | seems like you really thought about it all then | 10:56 |
@sonney2k | I would have just made the strict assumption - no messing with inputs - | 10:56 |
@sonney2k | I think you will have to give me a short 'tutorial' once you are done | 11:00 |
-!- blackburn [~blackburn@188.168.3.98] has quit [Quit: Leaving.] | 11:02 | |
-!- blackburn [~blackburn@188.168.4.202] has joined #shogun | 13:40 | |
-!- blackburn [~blackburn@188.168.4.202] has quit [Ping timeout: 240 seconds] | 13:46 | |
@sonney2k | heiko, what's up in the jungle? | 14:11 |
heiko | sonney2k, its horrible | 14:11 |
heiko | every time i touch something, new error araise | 14:12 |
heiko | I can now serialize/deserialize SGVectors | 14:12 |
heiko | but the normal vectors do not work | 14:12 |
@sonney2k | heiko, wait the normal serialization stuff no longer works? | 14:13 |
heiko | The most frequent problem is the problem to build a TParameter isntance that can call load, from sratch, meaning that class variables are not attached | 14:13 |
heiko | yes it does, but migration does not | 14:13 |
heiko | normal serialization works all the time | 14:13 |
heiko | because it is done separately | 14:13 |
heiko | but migration of the old style vectors does not yet | 14:14 |
heiko | its really really complicated, so many side-effects to think about | 14:14 |
@sonney2k | so for nomal vectors you need extra treatment right? | 14:15 |
@sonney2k | like copy doulbe* and len | 14:15 |
heiko | currently i am getting mad because i just dont get it how to create a TParameter instance to load an old style vector | 14:15 |
heiko | yes | 14:15 |
@sonney2k | and same with matrix/string/... | 14:15 |
heiko | no even touched this yet | 14:15 |
heiko | only doing vector/matrix | 14:16 |
@sonney2k | heiko, how about just *always* returning SGVector etc instead of normal vectors? | 14:16 |
@sonney2k | I mean when you have to do migrations then SGVector exists and is a more clean structure to store things | 14:17 |
@sonney2k | so then you don't need special extra treatment and the person doing migrations knows that he will only get scalars or SG* structures | 14:17 |
heiko | I also thought of that some time ago but then I would have to modify things even in a deeper layer | 14:17 |
heiko | which then causes even more horrible problems | 14:17 |
heiko | I am getting invalid writes/reads all the time | 14:18 |
@sonney2k | hmmhhh, what problems do you have in more deep layers? | 14:18 |
heiko | i would have to modify Tparameter::load for that (to store old style vectors in SGVector) | 14:19 |
serialhex | sonney2k: ok, i'll doo that and re-post it | 14:19 |
@sonney2k | serialhex, thanks | 14:41 |
@sonney2k | heiko, sorry had to answer some questions here | 14:41 |
@sonney2k | heiko, that sounds a lot simpler to me | 14:42 |
@sonney2k | ^ modifyingt Tparameter::load to optionally store old-style stuff in new-style vectors | 14:42 |
heiko | sonney2k, i got to go now, sorry | 14:58 |
@sonney2k | heiko, so what shall we do? | 14:58 |
heiko | i dont know | 14:59 |
heiko | this is simply not easy | 14:59 |
heiko | I dont know if it will be finished quick, I thought I could do it today | 14:59 |
heiko | but the current problem blocks it | 14:59 |
heiko | I pushed all code to my github account | 15:00 |
heiko | perhaps you can take a look | 15:00 |
heiko | here is my test program: | 15:00 |
heiko | http://pastebin.com/Xzxf0bKv | 15:00 |
@sonney2k | ok then I guess we will have to make a release w/o this - I don't know whether you will have time to work on this later on such that we can do a proper conversion | 15:00 |
heiko | and here the input file | 15:00 |
@sonney2k | later | 15:00 |
heiko | http://pastebin.com/y4xRpy0x | 15:00 |
@sonney2k | any thoughts? | 15:00 |
heiko | I will have tome to finish it, but the problem is that i move out of my flat on friday as I already told oyu | 15:01 |
heiko | I will have some time tomorrow in the morning but I really need to start on the flat | 15:01 |
@sonney2k | heiko, I know - I am just asking if at all | 15:01 |
-!- f-x [75c0d906@gateway/web/freenode/ip.117.192.217.6] has joined #shogun | 15:01 | |
@sonney2k | I mean I don't want to be an annoyance pushing you to do work you don't enjoy that much for too much longer | 15:01 |
@sonney2k | so I meant more like in the next weeks/months then | 15:02 |
heiko | yes, I know | 15:02 |
@sonney2k | I certainly prefer to keep you in the time than to push you beyond what is fun... - so it is really up to you | 15:03 |
@sonney2k | heiko, but if you commit doing it - it would certainly be great - but I don't know 1-2 months from now on ETA I guess then? | 15:03 |
heiko | yes something like that | 15:04 |
heiko | next two months are full | 15:04 |
heiko | but again, I think it will work soon, the conecept is there | 15:04 |
heiko | the problems are of technical nature | 15:04 |
heiko | lets do it like this: | 15:04 |
heiko | i will try to finish it in the next days, i have some hours per day | 15:05 |
heiko | and if it works, its great. | 15:05 |
@sonney2k | that sounds more like never then - the problem with this is that you really need to be so focussed - deep inside the code that it doing it as a side thing is close to impossible | 15:05 |
@sonney2k | ok | 15:05 |
@sonney2k | if not? | 15:05 |
heiko | if not, either you have a look, or I will do it later | 15:05 |
heiko | meaning starting again on it in one month | 15:05 |
@sonney2k | that is certainly fair | 15:06 |
@sonney2k | so we will try to do the tests w/o it and just create new regression tests with the new release | 15:06 |
@sonney2k | then make a release and we will see how people react | 15:07 |
heiko | perhaps wait one more day before you start on it | 15:07 |
@sonney2k | well we wanted to release tomorrow right :) | 15:07 |
heiko | ok then ..mmh | 15:07 |
@sonney2k | then blackburn starts studying again | 15:07 |
@sonney2k | I will have a new job ... so it is all difficult for us then for the next few weeks | 15:07 |
heiko | its stupid: the migration basically works, scalars work, SGVectors do work, only the normal vectors do not | 15:08 |
@sonney2k | which is the only thing we need to migrate | 15:08 |
heiko | But I will defenetly not finish it today | 15:08 |
@sonney2k | yeah I mean I could not foresee that it takes so long and that you were away most of last week | 15:09 |
@sonney2k | so it is like it is - | 15:10 |
heiko | Me too, thought it would be done in a week or so, but now I am working on it for two weeks, and the serialization bugs also took a lot of time | 15:10 |
@sonney2k | heiko, I am very happy that you intend to finish it and stay around later | 15:10 |
@sonney2k | and you've done a great job anyways | 15:10 |
@sonney2k | so we should be more in a party than fail mood :) | 15:11 |
heiko | thanks :) but a bit silly that this one does not work out | 15:11 |
heiko | true :) | 15:11 |
heiko | sonney2k, I got to go now, will be back tomorrow for some more talk and coding stuff | 15:11 |
heiko | bye | 15:12 |
@sonney2k | we will have another release w/ that feature soon | 15:12 |
@sonney2k | so don't worry thanks and bye | 15:12 |
-!- heiko [~heiko@134.91.52.222] has quit [Ping timeout: 258 seconds] | 15:16 | |
-!- blackburn [~blackburn@188.168.3.23] has joined #shogun | 15:25 | |
blackburn | sonney2k: any news? | 15:32 |
@sonney2k | blackburn, ys | 16:17 |
@sonney2k | yes | 16:17 |
@sonney2k | blackburn, we have to do w/o migrations for the release | 16:17 |
blackburn | oh | 16:20 |
blackburn | is heiko out of schedule? | 16:20 |
@sonney2k | blackburn, not out of schedule - it is much more difficult than we assumed | 16:21 |
@sonney2k | so we should try to check things manually now | 16:21 |
@sonney2k | that is use an old shogun version somewhere - run the python modular examples | 16:22 |
@sonney2k | and then compare the output | 16:22 |
@sonney2k | with what we get when we run this on some new machine | 16:22 |
@sonney2k | err shogun | 16:22 |
blackburn | hmm okay I see | 16:22 |
blackburn | sonney2k: do you have time for that testing? | 16:26 |
@sonney2k | not now | 16:27 |
@sonney2k | could you start a few things | 16:27 |
@sonney2k | ? | 16:27 |
blackburn | yeah sure, that's why I'm asking | 16:27 |
@sonney2k | thanks so please start and document which examples you testes | 16:30 |
@sonney2k | did test | 16:30 |
@sonney2k | btw you could try to run tester.py | 16:30 |
@sonney2k | lets see how much works :) | 16:30 |
blackburn | sonney2k: looks like nothing works :D | 16:42 |
@sonney2k | yay! | 16:42 |
@sonney2k | congrats | 16:42 |
blackburn | compiling old shogun | 16:44 |
@sonney2k | blackburn, *arg* the old shogun example also doesn't work | 16:58 |
@sonney2k | err the ocr one I meant | 16:58 |
CIA-3 | shogun: Soeren Sonnenburg master * rdc68e4f / (2 files): fix two more 'problematic' ruby examples - http://git.io/K_UBow | 17:48 |
CIA-3 | shogun: Soeren Sonnenburg master * r6f9f139 / src/shogun/io/BinaryFile.cpp : improve error output in binary file - http://git.io/P4d2iQ | 17:48 |
CIA-3 | shogun: Soeren Sonnenburg master * r4ecaa34 / (2 files): remove obsolete multiboost - http://git.io/QatMIw | 17:48 |
CIA-3 | shogun: Soeren Sonnenburg master * r312e97b / src/shogun/machine/KernelMachine.h : | 17:48 |
CIA-3 | shogun: Fix a bug in kernel machine introduced in SGVector transition | 17:48 |
CIA-3 | shogun: m_alpha has its own length and should be set and used independently from | 17:48 |
CIA-3 | shogun: m_svs otherwise serialization fails - http://git.io/HdbVAg | 17:48 |
CIA-3 | shogun: Soeren Sonnenburg master * rf0b3121 / (applications/ocr/Ai.py applications/ocr/data/ocr.svm.gz): fix ocr example and data for new shogun - http://git.io/gaA-Bg | 17:48 |
@sonney2k | blackburn, yay! progress | 17:49 |
CIA-3 | shogun: Soeren Sonnenburg master * r5a85eb5 / (2 files): fix sailfish application - http://git.io/1UfSnQ | 17:57 |
blackburn | sonney2k: sorry had to go | 18:02 |
blackburn | so are all applications operational now? | 18:03 |
CIA-3 | shogun: Soeren Sonnenburg master * rfaf1d9c / applications/vw/Makefile : add makefile for vw - http://git.io/KpTA2Q | 18:04 |
@sonney2k | blackburn, looks like yes | 18:05 |
blackburn | sonney2k: oh we got nice email with suggestion to implement some thing | 18:05 |
@sonney2k | what's missing now is check tests and make release script / debian package | 18:05 |
@sonney2k | blackburn, I would suggest you check the examples and I do the release scripts | 18:05 |
@sonney2k | blackburn, when ? | 18:06 |
@sonney2k | from whom | 18:06 |
blackburn | sonney2k: take a look on the mailing list | 18:06 |
@sonney2k | ahh ok | 18:06 |
@sonney2k | yes | 18:06 |
blackburn | sonney2k: will you answer? | 18:07 |
@sonney2k | nice - please reply that this is warmly welcome | 18:07 |
@sonney2k | can you do that? | 18:07 |
@sonney2k | just now | 18:07 |
blackburn | sonney2k: yeah okay | 18:07 |
@sonney2k | gtg | 18:07 |
@sonney2k | l8r | 18:07 |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 18:27 | |
shogun-buildbot | build #171 of octave_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/octave_modular/builds/171 | 18:41 |
blackburn | I think there is no reason to check all distances/kernels | 18:57 |
blackburn | will check some of them now | 18:57 |
@sonney2k | blackburn, I think you can even script this or? | 19:02 |
@sonney2k | *food* | 19:02 |
blackburn | sonney2k: need your help to automate this :) | 19:02 |
blackburn | sonney2k: surprise! gaussian kernel changed result | 19:07 |
blackburn | distance too! | 19:09 |
blackburn | even euclidian distance changed?! | 19:12 |
blackburn | ah | 19:15 |
blackburn | I guess I know | 19:15 |
blackburn | sonney2k: our saved results are totally useless: we used wrong data reading | 19:17 |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 19:19 | |
blackburn | okay distances and kernels should be ok | 19:42 |
blackburn | we haven't changed any compute routines so I'm pretty sure all OK | 19:42 |
blackburn | I had some doubts about parallel things but it is working smoothly | 19:43 |
@sonney2k | blackburn, btw let me check if I regenerate the tests if they work or not | 19:57 |
@sonney2k | kernel machines were broken before my last commit | 19:57 |
@sonney2k | blackburn, please try the predict example in applications/ocr | 19:57 |
@sonney2k | it's cool :) | 19:57 |
blackburn | sonney2k: I tried it days before | 19:59 |
@sonney2k | you mean years? | 19:59 |
blackburn | sth like 0.3 year before hah | 19:59 |
@sonney2k | argh | 20:02 |
@sonney2k | new crashers | 20:02 |
blackburn | ? | 20:02 |
@sonney2k | try python generator.py in python_modular testsuite | 20:03 |
blackburn | uh | 20:06 |
blackburn | sonney2k: | 20:08 |
blackburn | ==22401== Access not within mapped region at address 0x0 | 20:08 |
blackburn | ==22401== at 0x9C14FF1: shogun::TParameter::save_ptype(shogun::CSerializableFile*, void const*, char const*) (Parameter.cpp:1987) | 20:08 |
blackburn | ==22401== by 0x9C1545F: shogun::TParameter::save_stype(shogun::CSerializableFile*, void const*, char const*) (Parameter.cpp:2054) | 20:08 |
blackburn | ==22401== by 0x9C15E1C: shogun::TParameter::save(shogun::CSerializableFile*, char const*) (Parameter.cpp:2193) | 20:08 |
@sonney2k | yeah and other issues... | 20:09 |
blackburn | nice | 20:09 |
@sonney2k | yeah isn't it. it will never be boring | 20:11 |
@sonney2k | blackburn, there are other valgrind errors before even | 20:17 |
@sonney2k | well actually it starts in the first file | 20:17 |
@sonney2k | classifier_averaged_perceptron_modular.py | 20:17 |
CIA-3 | shogun: Soeren Sonnenburg master * rc52988d / (20 files in 3 dirs): | 20:28 |
CIA-3 | shogun: Merge pull request #320 from serialhex/master | 20:28 |
CIA-3 | shogun: Many more ruby examples (+6 more commits...) - http://git.io/DtZoig | 20:28 |
@sonney2k | argh | 20:45 |
@sonney2k | I am not really making progress on this bug | 20:46 |
CIA-3 | shogun: Soeren Sonnenburg master * r03ee3f8 / (doc/pages/ModularTutorial.mainpage src/README.developer): mention apply() instead of classify() in tutorial and README - http://git.io/mK5npw | 20:58 |
@sonney2k | blackburn, are you making progress? | 21:02 |
shogun-buildbot | build #157 of ruby_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org:5347/builders/ruby_modular/builds/157 blamelist: sonne@debian.org, serialhex@gmail.com | 21:10 |
blackburn | sonney2k: sorry got disturbed there in dormitory | 21:25 |
blackburn | sonney2k: what's with ruby? | 21:27 |
shogun-buildbot | build #160 of python_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org:5347/builders/python_modular/builds/160 blamelist: sonne@debian.org, serialhex@gmail.com | 21:28 |
blackburn | sonney2k: ruby examples in python?! | 21:29 |
blackburn | serialhex: hey you probably shouldn't have to commit python examples modifications ;) | 21:40 |
CIA-3 | shogun: Sergey Lisitsyn master * r9580eb8 / (3 files): Restore wrong modified python modular examples - http://git.io/zqUzKg | 21:55 |
-!- f-x [75c0d906@gateway/web/freenode/ip.117.192.217.6] has quit [Ping timeout: 252 seconds] | 22:28 | |
blackburn | sonney2k: got fixes for these crashes | 22:31 |
shogun-buildbot | build #2 of nightly_default is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org:5347/builders/nightly_default/builds/2 | 22:35 |
shogun-buildbot | build #162 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/python_modular/builds/162 | 22:49 |
CIA-3 | shogun: Sergey Lisitsyn master * r21b24b3 / src/shogun/machine/DistanceMachine.cpp : Fixed segmentation fault error emerged with DistanceMachine serialization - http://git.io/0o1-XA | 23:00 |
shogun-buildbot | build #163 of python_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org:5347/builders/python_modular/builds/163 blamelist: sonne@debian.org | 23:09 |
blackburn | shogun-buildbot: what the fuck are you compiling? | 23:10 |
CIA-3 | shogun: Sergey Lisitsyn master * raf951c1 / (2 files): Added modelselection examples' description - http://git.io/2_CLeQ | 23:27 |
--- Log closed Wed Aug 31 00:00:06 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!