--- Log opened Thu Apr 04 00:00:19 2013 | ||
-!- heiko [~heiko@027fc813.bb.sky.com] has quit [Quit: Leaving.] | 00:10 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 01:25 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 02:02 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 02:19 | |
shogun-buildbot_ | build #343 of nightly_default is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/343 | 04:24 |
---|---|---|
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 08:35 | |
shogun-notifier- | shogun: Soeren Sonnenburg :master * 385bd08 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/385bd08015902ab080f0e0f4eea36db4a850b9af | 08:35 |
shogun-notifier- | shogun: really fail on buildbot when new tests fail | 08:35 |
shogun-notifier- | shogun: | 08:35 |
shogun-notifier- | shogun: - require new data version with updated test-data | 08:35 |
shogun-notifier- | shogun: - seed randomized examples for reproducible results | 08:35 |
shogun-notifier- | shogun: - let tester display list of failed examples | 08:35 |
@sonney2k | blackburn, that should do it... | 08:35 |
shogun-buildbot_ | build #995 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/995 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:17 |
-!- blackburn [~blackburn@83.234.54.216] has quit [Quit: Leaving.] | 09:50 | |
-!- n4nd0 [~nando@n186-p150.kthopen.kth.se] has joined #shogun | 10:09 | |
-!- lambday [b613104f@gateway/web/freenode/ip.182.19.16.79] has joined #shogun | 10:28 | |
-!- wiking [~wiking@huwico/staff/wiking] has quit [Ping timeout: 260 seconds] | 10:53 | |
-!- wiking [~wiking@info2k1.hu] has joined #shogun | 10:53 | |
-!- n4nd0 [~nando@n186-p150.kthopen.kth.se] has quit [Quit: leaving] | 11:09 | |
-!- n4nd0 [~test@n150-p177.kthopen.kth.se] has joined #shogun | 11:10 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 11:35 | |
-!- heiko [~heiko@nat-170-240.internal.eduroam.ucl.ac.uk] has joined #shogun | 11:37 | |
-!- blackburn [~lisitsin@mxs.kg.ru] has joined #shogun | 11:41 | |
blackburn | sonne|work: libsvr? | 11:42 |
blackburn | hah we never know what will fail | 11:42 |
heiko | blackburn: thats my fault | 11:43 |
heiko | what is failing? | 11:43 |
heiko | blackburn: ah I fucked up the data again :( | 11:45 |
blackburn | heiko: hah alright | 11:46 |
heiko | blackburn: where do I have to commit data again? which dir? | 11:46 |
blackburn | heiko: shogun/data | 11:46 |
blackburn | heiko: ah btw I need your comments on | 11:46 |
blackburn | https://github.com/shogun-toolbox/shogun/issues/959 | 11:47 |
blackburn | heiko: ah and one more thing | 11:48 |
blackburn | heiko: what do you think about the following syntax | 11:48 |
blackburn | heiko: svm.parameter("width").set(5.0); | 11:48 |
blackburn | OR | 11:49 |
blackburn | svm.parameter(Width).set(5.0); | 11:49 |
blackburn | the latter with enum could be easily implemented as I see | 11:50 |
heiko | blackburn: for what? | 11:50 |
heiko | in order to avoid getter/setter stuff? | 11:50 |
blackburn | heiko: yes | 11:50 |
heiko | very cool idea | 11:50 |
heiko | much easier | 11:50 |
heiko | better to implement etc | 11:50 |
blackburn | heiko: but not only for that | 11:50 |
heiko | I think we should use strings | 11:50 |
heiko | since do not have to mess with enums, | 11:51 |
heiko | thats always annoying in python | 11:51 |
blackburn | heiko: yes I like strings but enum is easier to implement | 11:51 |
heiko | blackburn: what means easier? for who? | 11:51 |
blackburn | heiko: for us and users | 11:51 |
heiko | lets do the thing thats easier for users | 11:51 |
blackburn | parameter(DoubleParameterEnum) would return the thing | 11:51 |
heiko | but why? | 11:51 |
blackburn | that can be set only with double | 11:51 |
blackburn | that's not that easy with string | 11:52 |
heiko | blackburn: my fear is that one forgets to define those enums | 11:52 |
blackburn | yeah I like strings too | 11:52 |
blackburn | you know it is an anti-pattern | 11:52 |
blackburn | but still | 11:52 |
heiko | blackburn: as for the model selection syntax, we really have to think about all possible cases | 11:52 |
heiko | this was a horrible pain to implement 2 years ago, since so many possible combinations | 11:53 |
heiko | so whatever we use, lets keep all combinations in mind | 11:53 |
blackburn | heiko: so you do hate getters too? | 11:53 |
heiko | in the c++ examples, I covered most of the possible parameter trees | 11:53 |
heiko | blackburn: yes totally | 11:53 |
blackburn | brother | 11:53 |
blackburn | :D | 11:53 |
heiko | why do we have the parameter framework? | 11:53 |
heiko | haha :D | 11:53 |
blackburn | heiko: my idea is to store all parameters in map | 11:54 |
heiko | blackburn, so your example in the issue are fine | 11:54 |
heiko | but problems is sub-parameters | 11:54 |
blackburn | heiko: like? | 11:54 |
heiko | like sonney2k 's exam?ple | 11:54 |
blackburn | no that's not a problem | 11:54 |
blackburn | let me show how would I do that | 11:55 |
heiko | blackburn: I like this syntax more | 11:55 |
heiko | since it is more clear what hjappens | 11:55 |
heiko | ok | 11:55 |
blackburn | svm.parameter("C").as(ValuesSet).from([2,4]) | 11:55 |
blackburn | powerkernel.parameter("distance").as(ValuesSet).from([minkowski]) | 11:56 |
blackburn | etc | 11:56 |
blackburn | heiko: which syntax do you like more? | 11:56 |
sonne|work | heiko: failing libsvr makes sense - please commit to data and then do git add data and commit from the main repo and while you are at it please also regenerate for kernel_local_alignment_string_modular.py - I've changed that too | 11:56 |
heiko | sonne|work: already did, I will add the kernel stuff | 11:57 |
lambday | heiko, blackburn : hi :) morning | 11:57 |
sonne|work | heiko: but did you git push to data | 11:57 |
sonne|work | ? | 11:57 |
blackburn | lambday: hey | 11:57 |
sonne|work | heiko: *and* to the src repo? | 11:57 |
lambday | heiko, blackburn : I made some changes in that commit.. | 11:57 |
blackburn | heiko: yeah you probably forgot to update link in shogun | 11:57 |
heiko | lambday: morning! | 11:58 |
blackburn | git add data | 11:58 |
lambday | now it uses amd for both older and newer eigen | 11:58 |
blackburn | shogun/data I mean | 11:58 |
blackburn | lambday: nice | 11:58 |
heiko | blackburn, what is the procedure to update data commits? sorry for being stupid but I thought I have to push to data and thats it? | 11:58 |
heiko | lambday: very good! | 11:58 |
blackburn | heiko: no, go to shogun main repo now | 11:58 |
lambday | heiko: tried various methods since last night... seems like extending simplicialcholesky to use some of its protected members was the easiest among all | 11:58 |
heiko | lambday: did you write a test for both cases? | 11:58 |
blackburn | and update data dir too | 11:58 |
sonne|work | heiko: do you want the git lecture? | 11:59 |
blackburn | heiko: data dir in *main* repo points to some 'revision' | 11:59 |
blackburn | you updated the data but now you need to update that pointer in shogun *main* repo | 11:59 |
sonne|work | heiko: git can also keep submodules in sync with the main shogun repo | 11:59 |
heiko | blackburn: I see | 11:59 |
lambday | heiko: ummm... only one test case is there... if shogun is configured with older eigen it will use older method, otherwise newer method... I tested with both and it gives same result | 11:59 |
sonne|work | heiko: that is you need to commit the revision number of data thing by doing git commit data | 12:00 |
sonne|work | from the main shogun repo | 12:00 |
sonne|work | and then push | 12:00 |
heiko | lambday: ok good then, I will merge soon! | 12:00 |
heiko | sonne|work: let me try | 12:00 |
sonne|work | heiko: now when you do git pull it will also update data to the required state | 12:00 |
sonne|work | that helps to keep test data *AND* src code in sync | 12:00 |
lambday | heiko: :) I'll start looking into the Sampling large sparse Gaussians issue then... | 12:02 |
heiko | sonne|work: so when I use the generator to generate a new test, git tells me no changes :( | 12:05 |
heiko | from both data and shogun | 12:05 |
sonne|work | heiko: for which .py files? | 12:05 |
sonne|work | and in data you are on main branch? | 12:06 |
sonne|work | master I mean | 12:06 |
heiko | yes master | 12:06 |
heiko | python generator.py ../../../examples/undocumented/python_modular/kernel_local_alignment_string_modular.py | 12:06 |
sonne|work | heiko: no just do | 12:06 |
sonne|work | generator.py kernel_local_alignment_string_modular.py | 12:06 |
heiko | was that changed recently? ... might just be my bad memory | 12:07 |
sonne|work | no was always like this | 12:07 |
heiko | sonne|work: still git status doesnt contain the new files | 12:07 |
sonne|work | heiko: and you are in the data/ folder? | 12:08 |
heiko | the data repo or the data folder of the tests=? | 12:08 |
sonne|work | shogun/data | 12:08 |
heiko | yes | 12:08 |
heiko | heiko@heiko-ThinkPad-T420:~/Desktop/shogun/data$ git status | 12:09 |
heiko | # On branch master | 12:09 |
heiko | nothing to commit (working directory clean) | 12:09 |
sonne|work | and on master branch? | 12:09 |
sonne|work | ok the please remove the file | 12:09 |
heiko | the new data file? | 12:09 |
sonne|work | data/tests/kernel_local_alignment_string_modular* | 12:10 |
sonne|work | and see if git sees that there is a change | 12:10 |
heiko | I am totally confused, I changed all these tests a while ago and had no problems with it | 12:11 |
sonne|work | so is there one now? | 12:11 |
sonne|work | if you remove that file? | 12:11 |
heiko | yes that is detected | 12:12 |
heiko | I create the new ones ... | 12:12 |
heiko | git status still tells me deleted | 12:12 |
heiko | the genereator does not create the file ... | 12:13 |
heiko | let me check how I call it | 12:13 |
sonne|work | python generator.py kernel_local_alignment_string_modular.py | 12:15 |
sonne|work | but you need *latest* shogun compiled and python modular installed | 12:16 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 12:19 | |
shogun-notifier- | shogun: lambday :master * 7ba436e / / (3 files): https://github.com/shogun-toolbox/shogun/commit/7ba436ec19ed7de4e95efe9fa0b70f30b18b576d | 12:19 |
shogun-notifier- | shogun: added log-det method for sparse matrices in CStatistics | 12:19 |
shogun-notifier- | shogun: Heiko Strathmann :master * dba5cb7 / / (3 files): https://github.com/shogun-toolbox/shogun/commit/dba5cb777caa18e0e0e9d3fe3ab85c8a714d2f3e | 12:19 |
shogun-notifier- | shogun: Merge pull request #961 from lambday/master | 12:19 |
shogun-notifier- | shogun: | 12:19 |
shogun-notifier- | shogun: Added log-determinant method for sparse matrices in CStatistics class | 12:19 |
heiko | sonne|work: that might be the reason, of course! | 12:20 |
heiko | sonne|work: ok I have pushed to the data repo, will now commit the data version and push | 12:23 |
heiko | lets see if it works | 12:23 |
shogun-buildbot_ | build #996 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/996 blamelist: Soeren Sonnenburg <sonne@debian.org> | 12:25 |
shogun-notifier- | shogun: Heiko Strathmann :master * d4f8bf8 / data: https://github.com/shogun-toolbox/shogun/commit/d4f8bf854b4e7527a268f83c6fa6177d538ed899 | 12:25 |
shogun-notifier- | shogun: new tests for libsvr and local alignment kernel | 12:25 |
shogun-notifier- | shogun: Heiko Strathmann :master * 7987f7c / data: https://github.com/shogun-toolbox/shogun/commit/7987f7c8feb783d27f2b077871a33db4bee45f6a | 12:25 |
shogun-notifier- | shogun: Merge pull request #964 from karlnapf/master | 12:25 |
shogun-notifier- | shogun: | 12:25 |
shogun-notifier- | shogun: new data version | 12:25 |
heiko | sonne|work, blackburn, that should be it right? | 12:25 |
heiko | lambday: I merged your code, nice work! | 12:26 |
lambday | heiko: thanks man.. :) | 12:26 |
heiko | lambday: yes, sampling the large Gaussians is very similar, you have to permute the matrix, use the cholesky to sample, and then permute back | 12:26 |
heiko | you might want to look into the Gaussian class | 12:26 |
heiko | since it already can sample | 12:26 |
heiko | but maybe we want another approach | 12:27 |
heiko | dont know yet, feel free to suggest | 12:27 |
lambday | heiko: as of now, CGaussian doesn't use SGSparseMatrix I guess.. let me check | 12:27 |
lambday | hmm | 12:27 |
heiko | lambday: exactly | 12:27 |
heiko | sonne|work: so the integration tests now make the buildbot fail, nice! | 12:28 |
heiko | sonne|work: what about adding python_modular to the travis at some point? at least one simple version | 12:28 |
blackburn | heiko: yes should be right | 12:29 |
heiko | and I also would like to see the warnings as errors for the travis (at least for libshogun and python) | 12:29 |
heiko | but thats quite restrictive :D | 12:29 |
heiko | anyway, have to do some work now ... | 12:29 |
lambday | heiko: by the way, we'll be needing that SGSparseMatrix to eigen's SparseMatrix again... what about we add this method somewhere? blackburn asked to talk to sonney2k regarding this.. | 12:30 |
heiko | lambday: ah yes, I remember | 12:30 |
heiko | it would be good to have tools for this, but yes, swig could cause trouble | 12:30 |
heiko | I dont know what might happen, ask sonney2k aka sonne|work | 12:31 |
lambday | heiko: hmmm... | 12:31 |
heiko | lambday: what do you think about the CGaussian class? | 12:31 |
lambday | heiko: I just had a look... | 12:32 |
heiko | lambday: I think its the wrong place | 12:33 |
heiko | these CDistribution classes have a different purpose | 12:33 |
lambday | heiko: for this, we need to add a sample method (like the one that's already there), right? the one that uses cblas | 12:33 |
lambday | heiko: what's the wrong? | 12:33 |
heiko | so maybe, we just add sample_gaussian to CStatistics? | 12:33 |
lambday | heiko: oh! | 12:33 |
heiko | so what about a method to sample a multivariate Gaussian in CStatistics? | 12:34 |
heiko | both for sparse and dense matrices | 12:34 |
heiko | you give mean and covariance matrix and a number n | 12:34 |
heiko | you get n samples from that Gaussian | 12:34 |
heiko | you could start again with dense to get a feeling for the method | 12:34 |
heiko | and then extend to sparse (permutation stuff again) | 12:34 |
lambday | heiko: that sounds good... | 12:35 |
heiko | for the sparse, you have to compute the chol of the permuted matrix, and have to permute the samples backwards | 12:35 |
heiko | in the unit test, you can sample a few thousand points and then compute the empirical covariance matrix (we have that in shogun) | 12:36 |
heiko | and then compare | 12:36 |
lambday | okay.. | 12:36 |
lambday | heiko: let me see... will be shooting you with more stupid questions soon :D | 12:40 |
heiko | lambday: there are no stupid questions :D | 12:40 |
heiko | lambday: btw you could also replace the covariance method we have with an eigen3 call | 12:41 |
heiko | that would be useful | 12:41 |
lambday | covariance method in? | 12:43 |
lambday | CStatistics? | 12:44 |
shogun-buildbot_ | build #703 of cyg1 - libshogun is complete: Failure [failed configure] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/703 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 12:44 |
lambday | oh no! | 12:44 |
heiko | lambday: thats cyg, dont worry | 12:45 |
lambday | heiko: phew! | 12:45 |
heiko | http://www.shogun-toolbox.org/buildbot/tgrid | 12:45 |
lambday | hmm | 12:46 |
heiko | lambday: http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/997/steps/compile%20python_modular/logs/warnings%20%287%29 | 12:54 |
heiko | could you rename the variables so that swig doesnt do this col_ row_ underscore thing? | 12:54 |
heiko | and also please tell doxygen to ignore them | 12:55 |
heiko | this way, we get rid of warnings | 12:55 |
heiko | or maybe docunment them | 12:55 |
lambday | heiko: you mean rename col_ to col etc? | 12:57 |
heiko | lambday: swig renames col to col_ since col is reserved | 12:58 |
heiko | lambday: that is a bit annoying, but can be avoided via different names | 12:58 |
heiko | for example for SGVector, we use | 12:59 |
heiko | vlen for the length | 12:59 |
heiko | len is reserved | 12:59 |
lambday | ohkay.. | 12:59 |
heiko | just be creative | 12:59 |
heiko | and also tell doxygen to ignore or write comments with /** */ | 12:59 |
lambday | heiko: and how do I tell doxygen to ignore the "not documented" part | 12:59 |
heiko | #ifndef DOXYGEN_SHOULD_SKIP_THIS | 13:00 |
heiko | blackburn: right? ^ | 13:00 |
lambday | okay | 13:01 |
-!- blackburn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 13:02 | |
heiko | lambday: search the shogun code for this DOXYGEN_SOULD_SKIP_THIS to see how it works | 13:02 |
heiko | wow snowstorm | 13:02 |
heiko | london whether is really nice these days | 13:02 |
lambday | heiko: snowstrom ... I've never seen one :( | 13:03 |
heiko | lambday: not missing too much, its grey and cold ;) | 13:04 |
heiko | my heating wasnt working for two days early this week | 13:04 |
heiko | which means about 5 degrees inside the flat | 13:05 |
lambday | heiko: :D hard to imagine sitting here at 30 degree C | 13:05 |
lambday | heiko: oh shit! :D | 13:05 |
heiko | haha :) I remember sweating a lot when I was in mumbai | 13:05 |
lambday | heiko: you were in mumbai? when? | 13:05 |
heiko | lambday: yes about 2 years ago, I stayed a few days in colaba before traveling southwards | 13:06 |
heiko | went on a backpack holiday in south-west india for about 4 weeks | 13:06 |
heiko | quite different place to here ;) | 13:06 |
lambday | heiko: aah.. :D | 13:07 |
-!- blackburn [~lisitsin@mxs.kg.ru] has joined #shogun | 13:07 | |
lambday | and I'm dying to do so same in europe | 13:07 |
heiko | blackburn: #ifndef DOXYGEN_SHOULD_SKIP_THIS is the way to avoid doxygen warnings right? | 13:07 |
blackburn | heiko: yes | 13:07 |
heiko | lambday: go ahead :) | 13:07 |
lambday | okay I'm checking it then | 13:07 |
heiko | lambday: probably easiest way is through academia I guess - many Indian guys here at UCL for example | 13:08 |
heiko | lambday: I gotta do some work, will be back later today | 13:08 |
lambday | heiko: hmm... alright | 13:08 |
lambday | ciao man :) | 13:08 |
blackburn | lambday: discussing ways to leave india? :) | 13:09 |
lambday | blackburn: more of a trip :P | 13:09 |
-!- n4nd0 [~test@n150-p177.kthopen.kth.se] has quit [Ping timeout: 252 seconds] | 13:09 | |
lambday | could use some break from studies and do some other stuffs :D | 13:10 |
lambday | may be next year | 13:10 |
blackburn | lambday: what year of your studies it is? | 13:11 |
lambday | blackburn: doing pre-final year in masters | 13:11 |
blackburn | the same here hah | 13:14 |
lambday | blackburn: ahan... planning for phd? | 13:15 |
shogun-buildbot_ | build #997 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/997 | 13:16 |
blackburn | lambday: I don't know - I am heavily involved into kind of industry | 13:16 |
blackburn | and the only way to consider is phd in eu/usa not here | 13:16 |
lambday | blackburn: same here.. will think of phd if I get a chance abroad.. where are you from basically? | 13:18 |
blackburn | lambday: russia | 13:19 |
lambday | blackburn: ah, vodka :D | 13:20 |
blackburn | lambday: yes of course | 13:20 |
lambday | :D | 13:20 |
lambday | I started with it.. 8 years back.. never stopped :D | 13:22 |
blackburn | lambday: oh we use it instead of milk | 13:23 |
lambday | blackburn: lol man! :D | 13:23 |
shogun-buildbot_ | build #998 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/998 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 13:25 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has left #shogun ["Leaving"] | 13:39 | |
-!- n4nd0 [~test@n150-p177.kthopen.kth.se] has joined #shogun | 13:40 | |
-!- n4nd0 [~test@n150-p177.kthopen.kth.se] has quit [Client Quit] | 13:40 | |
-!- n4nd0 [~nando@n186-p150.kthopen.kth.se] has joined #shogun | 13:42 | |
sonne|work | heiko: did you git push? | 13:49 |
heiko | sonne|work yes, check github, both tests still fail :( | 13:55 |
sonne|work | heiko: but github didn't build that yet AFAIK | 13:55 |
heiko | ill force a rebuild | 13:56 |
sonne|work | heiko: https://github.com/shogun-toolbox/shogun/commit/dba5cb777caa18e0e0e9d3fe3ab85c8a714d2f3e | 13:56 |
sonne|work | that is the last commit | 13:56 |
heiko | https://github.com/shogun-toolbox/shogun/commits/master | 13:56 |
heiko | shows the data commit | 13:56 |
heiko | and in data there is also the new file | 13:56 |
heiko | The following tests failed! kernel_local_alignment_string_modular.py setting 1/2 kernel_local_alignment_string_modular.py setting 2/2 regression_libsvr_modular.py setting 1/2 regression_libsvr_modular.py setting 2/2 | 13:57 |
heiko | build 999 :) | 13:57 |
blackburn | http://jobs.ebaycareers.com/silicon-valley/student-and-new-grad/jobid3365811-intern-software-engineer-ebay-jobs | 14:09 |
blackburn | haha ebay wants something like coffin for liblinear | 14:09 |
heiko | blackburn: thats awesome! | 14:15 |
heiko | we will use that in shogun :) | 14:15 |
blackburn | heiko: someone should consider applying for that job | 14:16 |
blackburn | it looks like a week of work | 14:17 |
blackburn | (if one messed with shogun) | 14:17 |
heiko | maybe :) | 14:17 |
sonne|work | blackburn, heiko well shogun can already do that all :D | 14:17 |
heiko | I dont have time | 14:18 |
heiko | sonne|work haha :D | 14:18 |
heiko | we should tell them and ask for money | 14:18 |
blackburn | sonne|work: yeah that looks stupid to make *research* on that | 14:19 |
blackburn | nothing to research here | 14:19 |
blackburn | I am quite surprised that ebay doesn't know that? | 14:21 |
blackburn | how do they train classifiers? | 14:22 |
sonne|work | I guess some SGD | 14:22 |
n4nd0 | blackburn: sometimes I think that everything has already been discovered | 14:22 |
n4nd0 | we just have to actually find which person has already done what we aim for :D | 14:23 |
blackburn | n4nd0: yeah that's true | 14:23 |
shogun-buildbot_ | build #999 of deb3 - modular_interfaces is complete: Failure [failed test python_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/999 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com> | 14:26 |
sonne|work | heiko: could you please run tester.py locally with regression_libsvr_modular.py and kernel_local_alignment_string_modular.py ? | 14:29 |
sonne|work | ^stuff still seems to fail on the bb | 14:30 |
heiko | sonne|work, works locally | 14:51 |
sonne|work | excellent! | 14:57 |
-!- blackburn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 15:08 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 15:25 | |
-!- kakashi_ [~kakashi_@nltk/kakashi] has quit [Ping timeout: 245 seconds] | 15:33 | |
lambday | heiko: let me know if I got this right - first I generate n random samples X, using CMath::randn_double(), then I compute the lower cholesky factor, L of the covariance matrix C, then I use it to compute Y = L * X, and then add mean to each of these to get the actual n samples.. | 15:37 |
heiko | lambday: yes | 15:38 |
heiko | thats how it works | 15:38 |
heiko | you will have to add the mean afterwards | 15:38 |
heiko | lambday: oh wait | 15:39 |
heiko | in fact, you have to create vectors from N(0,I) - so each element with randn | 15:39 |
heiko | and then n of those vectors | 15:39 |
heiko | what you can do is just to create a DxN matrix and fill all the elements with randn | 15:40 |
heiko | and be careful about transposes since they sometimes mess up things, you should check for all this in the unit tests, but thats the way in pronciple | 15:40 |
lambday | heiko: yes... hmm | 15:40 |
heiko | sonne|work, so works locally, got it commited, but fails on bb?? | 15:40 |
lambday | heiko: btw I fixed the doxygen warnings.. renamed the variables as well.. | 15:41 |
lambday | why is build failing? | 15:42 |
heiko | lambday: nice, would you send a PR just with that? | 15:42 |
heiko | lambday: my fault, dont worry | 15:42 |
lambday | heiko: hmm.. okay I am sending a PR | 15:42 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 15:46 | |
shogun-notifier- | shogun: lambday :master * 1873193 / src/shogun/mathematics/Statistics.h: https://github.com/shogun-toolbox/shogun/commit/1873193a5b005c04b0feec5705836af7a556cb76 | 15:46 |
shogun-notifier- | shogun: fixed doxygen warnings in CStatistics | 15:46 |
shogun-notifier- | shogun: Heiko Strathmann :master * 4c78dc8 / src/shogun/mathematics/Statistics.h: https://github.com/shogun-toolbox/shogun/commit/4c78dc81fc58a81592975015b4d1b6024013408a | 15:46 |
shogun-notifier- | shogun: Merge pull request #965 from lambday/master | 15:46 |
shogun-notifier- | shogun: | 15:46 |
shogun-notifier- | shogun: fixed doxygen warnings in CStatistics, renamed a few variables | 15:46 |
-!- n4nd0 [~nando@n186-p150.kthopen.kth.se] has quit [Quit: leaving] | 15:51 | |
lambday | heiko: one more thing... say, we empirically calculate the covariance matrix of the samples, how do we compare it with the original one in the unit test? | 15:53 |
heiko | make sure the norm of the difference is small | 15:53 |
heiko | "sufficiently small" | 15:53 |
heiko | same for the mean | 15:53 |
lambday | ohkay... got it | 15:54 |
lambday | hmmm | 15:54 |
-!- lambday_ [0e8b614f@gateway/web/freenode/ip.14.139.97.79] has joined #shogun | 15:58 | |
-!- lambday [b613104f@gateway/web/freenode/ip.182.19.16.79] has quit [Ping timeout: 245 seconds] | 15:58 | |
shogun-buildbot_ | build #704 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/cyg1%20-%20libshogun/builds/704 | 16:03 |
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has quit [Ping timeout: 260 seconds] | 16:04 | |
@sonney2k | heiko, let me check if the stuff fails here too... | 16:11 |
heiko | sonney2k: ok | 16:11 |
-!- sumit [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 16:11 | |
@sonney2k | heiko, local alignment kernel works fine here | 16:12 |
heiko | and libsvr? | 16:13 |
heiko | sonney2k: I changed that too | 16:13 |
@sonney2k | exception - what did you do? | 16:13 |
@sonney2k | can't we have two libsvr examples instead of one? | 16:14 |
@sonney2k | I mean one for epsilon / C and one for C & nu? | 16:14 |
@sonney2k | heiko, ^ | 16:18 |
heiko | ah sorry | 16:18 |
heiko | sonney2k: I changed the example to do both | 16:19 |
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has joined #shogun | 16:19 | |
heiko | sonney2k: dont get this, but its safe to overwrite -- I added unit tests which ensure that libsvr does the correct thing for epsilon and nu (compared against libsvm output) | 16:19 |
heiko | so just create a new test file | 16:20 |
heiko | dont know why mine doesnt work | 16:20 |
@sonney2k | heiko, no I meant it might be worthwhile to have each method in a separate file but IDK | 16:21 |
heiko | sonney2k: why? | 16:21 |
heiko | its just one line difference | 16:21 |
heiko | and the unit tests should tell us which one of those fail | 16:21 |
heiko | but we can also change it | 16:21 |
@sonney2k | heiko, looks like you didn't update the svr data | 16:23 |
@sonney2k | at least the output is once len == 100 and once 120 | 16:24 |
-!- blackburn [~lisitsin@mxs.kg.ru] has joined #shogun | 16:24 | |
heiko | sonney2k: https://github.com/shogun-toolbox/shogun-data/commit/3e9b26391c67e6d4291b3a5138d7a403778624c7 | 16:25 |
heiko | but I dont know, I always struggle with this | 16:25 |
heiko | could you add a new one? was working before and I only changed example a bit | 16:25 |
@sonney2k | heiko, but did you git add data *afterwards* | 16:25 |
heiko | yes | 16:26 |
heiko | apparently not, but I thought I did | 16:26 |
@sonney2k | no you didn't | 16:26 |
heiko | I committed this yesterday | 16:26 |
heiko | and updated the data today | 16:26 |
@sonney2k | let me update the data ptr | 16:27 |
heiko | cool thanks! | 16:27 |
@sonney2k | heiko, btw both works here just fine | 16:27 |
shogun-buildbot_ | build #1000 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/1000 | 16:41 |
shogun-buildbot_ | build #1001 of deb3 - modular_interfaces is complete: Failure [failed test ruby_modular] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb3%20-%20modular_interfaces/builds/1001 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 16:47 |
-!- kakashi_ [~kakashi_@nltk/kakashi] has joined #shogun | 16:53 | |
-!- sumit [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Quit: Page closed] | 16:55 | |
-!- tp199911 [~thomas@109.129.173.247] has joined #shogun | 16:57 | |
-!- tp199911 is now known as ZeThomad | 16:57 | |
-!- ZeThomad is now known as ZeThomas | 16:57 | |
ZeThomas | hey sonney2k, blackburn | 17:34 |
ZeThomas | I get weird stuff in my code | 17:35 |
ZeThomas | it just almost immediately exits with code 139, no errors | 17:35 |
ZeThomas | here's my code: http://pastebin.com/567TrTnx | 17:36 |
ZeThomas | I had the same thing when I did a similar thing with a CustomKernel, but that got solved by no longer subclassing CustomKernel, as sonney2k pointed out | 17:39 |
ZeThomas | But DirectorKernels are made for being subclassed, no? Or is there another way? | 17:39 |
blackburn | ZeThomas: hey | 17:49 |
blackburn | let me take a look | 17:49 |
blackburn | ZeThomas: sorry have to go - will take a look tonight | 17:52 |
-!- blackburn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 17:52 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:11 | |
ZeThomas | weird... I have implemented the exact same functionality in 2 different kernel setups, and yet the svm gives me different results... | 18:33 |
ZeThomas | an F1-score on average of 3% more when using the CustomKernel that does the exact same thing as a LinearKernel with big feature vectors | 18:36 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 256 seconds] | 18:37 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:41 | |
shogun-buildbot_ | build #1002 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/1002 | 18:42 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 18:46 | |
-!- shogun-toolbox [~shogun@7nn.de] has quit [Ping timeout: 256 seconds] | 19:01 | |
--- Log closed Thu Apr 04 19:01:21 2013 | ||
--- Log opened Thu Apr 04 19:01:29 2013 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 19:01 | |
-!- Irssi: #shogun: Total of 12 nicks [1 ops, 0 halfops, 0 voices, 11 normal] | 19:01 | |
-!- Irssi: Join to #shogun was synced in 6 secs | 19:01 | |
-!- lambday_ [0e8b614f@gateway/web/freenode/ip.14.139.97.79] has quit [Ping timeout: 245 seconds] | 19:24 | |
@sonney2k | ZeThomas, try to use a lower svm epsilon | 19:52 |
@sonney2k | say 1e-5 | 19:52 |
@sonney2k | ZeThomas, a custom kernel internally uses just floating point precision while a standard kernel uses double | 19:52 |
@sonney2k | that could alreayd explain it | 19:53 |
ZeThomas | sonney2k, but that implies that the less precise svm gives more precise predictions? | 19:54 |
@sonney2k | ZeThomas, sure why not | 19:54 |
ZeThomas | how would you explain that? | 19:54 |
@sonney2k | ZeThomas, SVM just optimizes to eps precision | 19:54 |
@sonney2k | so choice of eps / C / and floating point precison all matter | 19:55 |
ZeThomas | oh, so by lowering the eps in the LinearKernel, my results will be better aligned? | 19:55 |
ZeThomas | or in the CustomKernel? | 19:56 |
@sonney2k | heiko, looks like you did commit data after all - we have one green happy bb! | 19:56 |
@sonney2k | ZeThomas, both | 19:56 |
heiko | sonney2k: yes build 1002 worked :) | 19:56 |
@sonney2k | results should get more similar | 19:56 |
heiko | sonney2k: do you know what this weird warning in python modular means? | 19:57 |
@sonney2k | heiko, yes | 19:57 |
@sonney2k | some scipy module loading a .mat file | 19:57 |
@sonney2k | and it looks like it is not specifying some option | 19:58 |
heiko | ok | 19:58 |
@sonney2k | which is supposed to be changed in future versions | 19:58 |
@sonney2k | let me fix it | 19:58 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 245 seconds] | 20:01 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 20:04 | |
shogun-notifier- | shogun: Soeren Sonnenburg :master * d10e022 / examples/undocumented/python_modular/ (3 files): https://github.com/shogun-toolbox/shogun/commit/d10e022a23a26b51e73a217c5014f3cf8d879ee6 | 20:04 |
shogun-notifier- | shogun: fix a the loadmat warning | 20:04 |
@sonney2k | heiko, do you have git-flow available as a package? | 20:04 |
@sonney2k | I mean could you please install it | 20:04 |
@sonney2k | ? | 20:04 |
@sonney2k | heiko, now that we have everything 'green' we should switch to the new dev model | 20:05 |
@sonney2k | so it would be good taht you have this thing around too | 20:05 |
-!- ZeThomas [~thomas@109.129.173.247] has quit [Ping timeout: 258 seconds] | 20:16 | |
heiko | sonney2k: gitflow? | 20:19 |
@sonney2k | heiko, you are on ubuntu right? | 20:19 |
heiko | installed | 20:19 |
@sonney2k | heiko, http://labs.grupow.com/blog/2011/07/05/getting-started-with-git-flow | 20:19 |
@sonney2k | it is this scheme http://nvie.com/posts/a-successful-git-branching-model/ | 20:19 |
heiko | ah nice I see | 20:19 |
@sonney2k | for each feature we work on feature branches | 20:19 |
@sonney2k | *directly*( | 20:20 |
@sonney2k | and then merge stuff down to the development branch | 20:20 |
heiko | yes thats good | 20:20 |
@sonney2k | git flow helps us doing that all easily | 20:20 |
heiko | nice | 20:20 |
heiko | I will check it out ... | 20:20 |
@sonney2k | heiko, jsut type git flow help | 20:20 |
heiko | sonney2k: I will check it out later, currently busy with something, maybe lets do the switch on the weekend or next week? | 20:21 |
@sonney2k | heiko, ASAP I would say | 20:22 |
@sonney2k | when wiking/blackburn are around | 20:22 |
heiko | sonney2k: agreed | 20:22 |
heiko | lets just wait until everyone is around and then do it spontaneously | 20:22 |
@sonney2k | master will become less polluted | 20:22 |
heiko | I will be around on the weekend | 20:23 |
heiko | sonney2k: yes, that will be very useful | 20:23 |
heiko | I am also keen on activating python on travis, will also help a lot | 20:23 |
@sonney2k | heiko, well most important is to have our buildbot infrasturcture | 20:30 |
heiko | sonney2k: I am keen on this since it prevents us from being lazy | 20:30 |
@sonney2k | travis is not that helpful IMHO | 20:30 |
heiko | so warnings/errors dont add up | 20:30 |
@sonney2k | it is doing lots of false stuff | 20:30 |
heiko | you only merge if it succeeds | 20:30 |
heiko | since we sometimes ignore new warnings in merges | 20:31 |
heiko | same for new errors | 20:31 |
heiko | but agree buildbot more useful | 20:31 |
heiko | but travis is also very useful for gsoc | 20:31 |
heiko | when you dont write code on your own | 20:31 |
@sonney2k | heiko, well I am pretty happy with the infrastructure now | 20:36 |
heiko | sonney, yes its getting better and better :) | 20:36 |
@sonney2k | heiko, we had all these integration tests last year too but we didn't turn them on | 20:36 |
@sonney2k | in the buildbot | 20:36 |
@sonney2k | so that is why so much broke | 20:36 |
@sonney2k | now that we have everything in place it should be harder(tm) to break stuff | 20:37 |
@sonney2k | in addition, lots of shogun's internals are more developed now so we won't have huge changes like we had the gsoc's back | 20:37 |
heiko | sonney2k, yes, agreed. I really like the approach of making things that work waterproof | 20:43 |
heiko | so whenever something breaks, we will notice | 20:43 |
heiko | I guess next step is to improve documentation :) | 20:43 |
@sonney2k | heiko, for git flow watch this http://buildamodule.com/video/change-management-and-version-control-deploying-releases-features-and-fixes-with-git-how-to-use-a-scalable-git-branching-model-called-gitflow | 20:46 |
@sonney2k | heiko, and then this how to do it with git flow http://labs.grupow.com/blog/2011/07/05/getting-started-with-git-flow | 20:51 |
@sonney2k | it is really easy... | 20:51 |
@sonney2k | heiko, so I think we can switch right away | 20:51 |
-!- blackburn [~blackburn@109.226.90.103] has joined #shogun | 20:56 | |
@sonney2k | blackburn, we are greeeen! | 20:56 |
blackburn | sonney2k: heh nice | 20:56 |
@sonney2k | blackburn, lets switch to git flow ! | 20:56 |
@sonney2k | blackburn, do you already have git-flow installed? | 20:56 |
blackburn | sonney2k: no | 20:56 |
@sonney2k | blackburn, then please apt-get install it | 20:56 |
blackburn | sonney2k: I didn't know about that tool (ashamed) | 20:57 |
@sonney2k | blackburn, read here <sonney2k> heiko, for git flow watch this http://buildamodule.com/video/change-management-and-version-control-deploying-releases-features-and-fixes-with-git-how-to-use-a-scalable-git-branching-model-called-gitflow | 20:57 |
@sonney2k | <sonney2k> heiko, and then this how to do it with git flow http://labs.grupow.com/blog/2011/07/05/getting-started-with-git-flow | 20:57 |
blackburn | sonney2k: alrigh | 20:58 |
heiko | guys, sorry I cannot stay much longer today | 20:58 |
@sonney2k | heiko, no worries we will keep you updated | 20:58 |
blackburn | heiko: you won't forgiven | 20:59 |
blackburn | be* | 20:59 |
@sonney2k | blackburn, what I don't know is whether everyone has to run git flow init | 20:59 |
blackburn | sonney2k: what does it do exactly? | 20:59 |
@sonney2k | blackburn, it sets up the git flow layout | 20:59 |
blackburn | *exactly* which commands? | 21:00 |
@sonney2k | with feature branches, development branch main (aka master) branch etc | 21:00 |
@sonney2k | git flow init -d | 21:00 |
blackburn | checking sources | 21:00 |
@sonney2k | blackburn, I would very much love to go with the defaults | 21:00 |
blackburn | sonney2k: I think it is enough to do init only once | 21:01 |
blackburn | according to what I see in sources of gitflow | 21:01 |
@sonney2k | blackburn, you mean either of us can do it? | 21:01 |
blackburn | sonney2k: yes probably - it creates branches | 21:01 |
@sonney2k | blackburn, do you have local branches except master on your clone? | 21:01 |
blackburn | no | 21:02 |
blackburn | sonney2k: can we drop some branches btw? | 21:02 |
@sonney2k | blackburn, I mean on the shogun toolbox clone | 21:02 |
blackburn | sonney2k: yes I got you - I don't have any local branches | 21:02 |
@sonney2k | not shogun clone under your github account | 21:02 |
@sonney2k | OK | 21:02 |
@sonney2k | blackburn, then please do git flow init -d in that | 21:03 |
blackburn | sonney2k: I am concerned with some stuff here | 21:03 |
@sonney2k | shoot | 21:03 |
blackburn | sonney2k: there is such thing as feature publish | 21:04 |
blackburn | I think we are not going to use remotes right? | 21:04 |
blackburn | except main remore | 21:04 |
@sonney2k | blackburn, maybe not but if git flow creates remote branches for features why not | 21:05 |
@sonney2k | it doesn't hurt | 21:05 |
blackburn | sonney2k: okay let me clean my repo and init | 21:06 |
blackburn | sonney2k: ahh I was bored in bus and came with another api thing | 21:07 |
blackburn | sonney2k: data.features(1,20).disable() | 21:07 |
@sonney2k | blackburn, actually if I understand git correctly I could delete a local branch | 21:07 |
@sonney2k | while keeping the remote branch | 21:07 |
blackburn | sonney2k: yes | 21:07 |
blackburn | why not | 21:07 |
@sonney2k | because I have this stupid shogun_2.1.0 branch here | 21:08 |
@sonney2k | but I dont' need it | 21:08 |
blackburn | but it will appear on git pull | 21:08 |
blackburn | again | 21:08 |
@sonney2k | I only needed it to push to our 3 remotes | 21:08 |
@sonney2k | blackburn, no only when I explicitly pull it or? | 21:09 |
* sonney2k tries | 21:09 | |
blackburn | sonney2k: if you do just 'git pull' | 21:09 |
blackburn | not specifying any branch | 21:09 |
blackburn | it will fetch it for you again I think | 21:09 |
@sonney2k | git branch -d shogun_2.1 | 21:09 |
@sonney2k | error: The branch 'shogun_2.1' is not fully merged. | 21:09 |
@sonney2k | If you are sure you want to delete it, run 'git branch -D shogun_2.1'. | 21:09 |
@sonney2k | ugh | 21:09 |
blackburn | I did git flow init -d | 21:10 |
@sonney2k | blackburn, did that go through? | 21:10 |
@sonney2k | what branches do you have now? | 21:10 |
@sonney2k | git branch | 21:10 |
blackburn | Production and integration branches should differ. | 21:10 |
@sonney2k | haha | 21:10 |
blackburn | I think defaults are not ok for us | 21:10 |
blackburn | okay where do we develop *next release*? | 21:10 |
blackburn | sonney2k: unstable? | 21:10 |
@sonney2k | master | 21:11 |
@sonney2k | or develop? | 21:11 |
@sonney2k | I am confused by the wording | 21:11 |
blackburn | sonney2k: lets get back to the scheme | 21:11 |
blackburn | sonney2k: http://habrastorage.org/storage/4bf7e68c/49e29c35/3a01bd6b/782a1be3.png | 21:12 |
@sonney2k | develop then | 21:12 |
blackburn | sonney2k: if my understanding is correct - master is to store tags of releases | 21:12 |
blackburn | develop in what we do for next release | 21:13 |
blackburn | uhm I am confused | 21:14 |
@sonney2k | yeah I agree | 21:14 |
@sonney2k | develop == for next release | 21:14 |
blackburn | sonney2k: it didn't create anything | 21:14 |
blackburn | how does it work? | 21:14 |
@sonney2k | blackburn, did it run through? | 21:15 |
blackburn | yes | 21:15 |
@sonney2k | I mean did you answer all the questions? | 21:15 |
blackburn | sonney2k: I told it to use 'dev' | 21:15 |
@sonney2k | 5 or so | 21:15 |
@sonney2k | and then | 21:15 |
@sonney2k | ? | 21:15 |
blackburn | I kept feature/release/etc by default | 21:15 |
blackburn | does it store some config for that? | 21:15 |
@sonney2k | maybe it is a local thing after all? | 21:15 |
@sonney2k | looks like it | 21:15 |
@sonney2k | git config -l | 21:16 |
@sonney2k | it should list gitflow | 21:16 |
@sonney2k | then let me run git flow init too | 21:16 |
blackburn | ohh yes | 21:16 |
blackburn | I see | 21:16 |
@sonney2k | blackburn, so we use dev right? | 21:17 |
@sonney2k | not devel? | 21:17 |
blackburn | sonney2k: it requires existent branch | 21:17 |
@sonney2k | blackburn, I am asking because we have some dev branch crurently | 21:17 |
blackburn | sonney2k: lets kill dev | 21:17 |
blackburn | and create it once again | 21:17 |
@sonney2k | and it is urgs | 21:17 |
@sonney2k | OK | 21:17 |
@sonney2k | please kill it then | 21:17 |
@sonney2k | so we will have master/dev/feature/hotfix/release right? | 21:18 |
blackburn | sonney2k: lets call it develop | 21:18 |
@sonney2k | yes | 21:18 |
@sonney2k | lets do it | 21:18 |
blackburn | to not confuse with an old one | 21:18 |
blackburn | sonney2k: oh I start to realize the power of that thing | 21:19 |
blackburn | sonney2k: merging of hotfixes to releases sounds powerful | 21:20 |
@sonney2k | blackburn, please prefix tags with shogun_ | 21:20 |
blackburn | sonney2k: ok | 21:21 |
blackburn | sonney2k: can't we use only one git remote? | 21:21 |
blackburn | sonney2k: may be some cron script that pushes it to sg and google | 21:21 |
blackburn | I am afraid of breaking something if they go out of sync | 21:22 |
blackburn | sonney2k: pull | 21:23 |
@sonney2k | blackburn, probably - if you have time - please experiment with sth | 21:23 |
@sonney2k | blackburn, so do I have to run git flow too? | 21:23 |
blackburn | sonney2k: yes | 21:23 |
@sonney2k | or shall I just git pull -a now? | 21:23 |
blackburn | sonney2k: git pull to fetch new develop branch | 21:24 |
blackburn | and git flow init -f to select it for now | 21:24 |
blackburn | sonney2k: | 21:25 |
blackburn | gitflow.branch.master=master | 21:25 |
blackburn | gitflow.branch.develop=develop | 21:25 |
blackburn | gitflow.prefix.feature=feature/ | 21:25 |
blackburn | gitflow.prefix.release=release/ | 21:25 |
blackburn | gitflow.prefix.hotfix=hotfix/ | 21:25 |
blackburn | gitflow.prefix.support=support/ | 21:25 |
blackburn | gitflow.prefix.versiontag=shogun_ | 21:25 |
blackburn | sonney2k: ^ that's how my git flwo conf looks like | 21:25 |
-!- heiko [~heiko@nat-170-240.internal.eduroam.ucl.ac.uk] has left #shogun [] | 21:28 | |
@sonney2k | blackburn, so now we can just do git flow feature foo | 21:28 |
@sonney2k | work on this finish this | 21:28 |
@sonney2k | and done? | 21:29 |
@sonney2k | I am not sure how the push/pull stuff works | 21:29 |
@sonney2k | and how do we work with students from now on? | 21:29 |
@sonney2k | they all send PR's to dev right? | 21:29 |
blackburn | sonney2k: yes | 21:29 |
blackburn | sonney2k: from feature/something | 21:29 |
@sonney2k | and they work on their own feature branches right? | 21:29 |
@sonney2k | ok | 21:29 |
blackburn | I am not sure about pushing | 21:30 |
@sonney2k | ohh I like it | 21:30 |
@sonney2k | when we want to do a release | 21:30 |
@sonney2k | we can just start a release branch | 21:30 |
blackburn | yes | 21:30 |
@sonney2k | and then do all the fixing in there | 21:30 |
@sonney2k | while we can still go wild in dev | 21:30 |
blackburn | sonney2k: I am going to check the pushing thing | 21:31 |
@sonney2k | blackburn, I think you just do git push on the branch you are on to the branch you want to push | 21:32 |
@sonney2k | to | 21:32 |
blackburn | sonney2k: okay when you git flow feature start | 21:32 |
blackburn | it creates a branch | 21:32 |
@sonney2k | so it changed FETCH_HEAD to develop here | 21:32 |
blackburn | you commit some things | 21:32 |
@sonney2k | yeah but that is local right? | 21:32 |
blackburn | yes sure | 21:32 |
@sonney2k | so only when you wnat to work with sb on this | 21:32 |
@sonney2k | you push it | 21:33 |
blackburn | git flow feature finish *merges and deletes* | 21:33 |
@sonney2k | to create a new remote branch | 21:33 |
@sonney2k | yes | 21:33 |
@sonney2k | so you have a single clean commit | 21:33 |
blackburn | sonney2k: only one commit? | 21:34 |
blackburn | does it stack it? | 21:34 |
@sonney2k | blackburn, I am only guessing | 21:35 |
blackburn | checking | 21:35 |
blackburn | sonney2k: no | 21:35 |
blackburn | it puts all commits | 21:36 |
blackburn | and merge commit too | 21:36 |
blackburn | it is possible to stack them too I think | 21:36 |
blackburn | with some patching may be | 21:36 |
@sonney2k | blackburn, the good thing is that you can do as many git reset <id> as you want | 21:38 |
@sonney2k | so you /can/ cleanup your local feature branch | 21:38 |
@sonney2k | to not have crappy / build breaking etc stuff in there | 21:38 |
blackburn | sonney2k: I'd like to have kinda selective rights for branches | 21:38 |
@sonney2k | yeah no one can push to master... | 21:39 |
@sonney2k | except you or os | 21:39 |
@sonney2k | so | 21:39 |
blackburn | sonney2k: then we would allow commits for everyone (in feature branches) | 21:39 |
@sonney2k | blackburn, well then our repo would be spammed | 21:40 |
blackburn | sonney2k: no, for students + us I mean | 21:40 |
@sonney2k | yes sure | 21:40 |
@sonney2k | blackburn, actually we don't need that for students - they can work on their own fork | 21:43 |
@sonney2k | and from there do the PRs | 21:43 |
@sonney2k | issue with git flow is that this doesn't go very well with github | 21:43 |
@sonney2k | so PR's are some kind of alternate concept | 21:43 |
@sonney2k | and other remote forks | 21:43 |
@sonney2k | blackburn, please announce on the mailinglist that we changed the dev model | 21:46 |
@sonney2k | and add the steps | 21:46 |
@sonney2k | maybe we need to add some page on our website where we document the steps | 21:47 |
@sonney2k | (or at least our README.developer) | 21:47 |
blackburn | sonney2k: ok | 21:50 |
@sonney2k | blackburn, what is left to do for that is - what do we do with the buildbot? | 21:50 |
@sonney2k | should it form now on work on develop? | 21:51 |
blackburn | sonney2k: yes as master is to be updated only a few times per year may be :) | 21:51 |
@sonney2k | and actually master will just be some state we had in develop / the release branch | 21:52 |
blackburn | yes | 21:53 |
@sonney2k | the only issue I see is that we would want to have a buildbot for release branches | 21:53 |
blackburn | sonney2k: why? | 21:53 |
@sonney2k | blackburn, tomorrow gsoc will be announced right? | 21:53 |
blackburn | sonney2k: oh I am afraid 7 | 21:53 |
blackburn | April 8: 19:00 UTCList of accepted mentoring organizations published on the Google Summer of Code 2013 site. | 21:54 |
@sonney2k | blackburn, because one directly commits to the release branch trying to stabilize things | 21:54 |
@sonney2k | puh | 21:54 |
@sonney2k | so we still have time | 21:54 |
blackburn | monday | 21:54 |
blackburn | sonney2k: for what? | 21:54 |
@sonney2k | blackburn, recall our todo? | 21:54 |
@sonney2k | 1) get build to work | 21:54 |
@sonney2k | 2) new dev model | 21:55 |
@sonney2k | 3) update website | 21:55 |
blackburn | sonney2k: oh this year gsoc is so sloooow | 21:55 |
@sonney2k | 4) announce workshop | 21:55 |
@sonney2k | ->then get accepted (please!!) | 21:55 |
blackburn | interim is 8-22 May | 21:55 |
blackburn | crazy! | 21:55 |
@sonney2k | blackburn, how was it last year | 21:55 |
blackburn | ohh month back or even more | 21:55 |
@sonney2k | I totally forgot... | 21:55 |
blackburn | sonney2k: it started officially in early may | 21:56 |
@sonney2k | yeah I remember it was quite some rush | 21:56 |
blackburn | sonney2k: this time it starts with summer and ends in september | 21:56 |
@sonney2k | I think they are trying to accomodate for how we study in europe | 21:56 |
blackburn | yes probably | 21:56 |
@sonney2k | vacations here are july-september | 21:57 |
blackburn | in usa as well? | 21:57 |
@sonney2k | no idea | 21:57 |
@sonney2k | blackburn, btw https://dl.dropbox.com/u/31480471/website-cbase.png | 21:58 |
@sonney2k | n4nd0 suggested to put the c-base workshop stuff on the long thing | 21:58 |
@sonney2k | and the gsoc announcement in the middle? | 21:58 |
@sonney2k | thoughts? | 21:58 |
blackburn | sonney2k: oh I totally dislike everything sorry :) | 21:59 |
blackburn | we should do something with it | 21:59 |
@sonney2k | what? | 21:59 |
@sonney2k | 100% parse error | 21:59 |
blackburn | sonney2k: haha I don't like whole design of the page | 22:00 |
@sonney2k | haha | 22:00 |
blackburn | sonney2k: soo I don't have any ideas | 22:01 |
blackburn | what this cbase thing is doing here? | 22:01 |
blackburn | sonney2k: I am mostly depressed about that box | 22:01 |
blackburn | it looks like 2002 or 2003 idk | 22:02 |
blackburn | 3d box | 22:02 |
@sonney2k | blackburn, which box? | 22:02 |
blackburn | left to shogun thing | 22:03 |
blackburn | with 'new' label | 22:03 |
blackburn | sonney2k: do you feel ok about it? | 22:03 |
@sonney2k | blackburn, sure I like it | 22:03 |
@sonney2k | the c-base logo - we want to announce the workshop | 22:03 |
blackburn | oh I don't know may be it is something wrong with me :) | 22:03 |
@sonney2k | so some kind of logo should be there | 22:04 |
@sonney2k | some graphics - text is too boring... | 22:04 |
blackburn | sonney2k: http://twitter.github.com/bootstrap/examples/carousel.html | 22:05 |
@sonney2k | blackburn, so you like this style? | 22:13 |
blackburn | sonney2k: which? | 22:13 |
@sonney2k | this bootstrap link above | 22:14 |
blackburn | oh yes it is awesome | 22:14 |
blackburn | bootstrap stuff looks very cool (for me) | 22:14 |
@sonney2k | it is wasting even more space than our theme though | 22:15 |
blackburn | sonney2k: are you tired with my crazy api things? I have some more to show you | 22:16 |
@sonney2k | blackburn, in any case switching themes is just a matter of someone changing the css | 22:16 |
@sonney2k | that's all | 22:16 |
blackburn | sonney2k: no with bootstrap classes should be quite different | 22:16 |
@sonney2k | blackburn, sure I am :D | 22:16 |
blackburn | sonney2k: | 22:17 |
@sonney2k | blackburn, why? | 22:17 |
blackburn | svm.train().now() | 22:17 |
blackburn | svm.train().meanwhile() | 22:17 |
blackburn | svm.train().later() | 22:17 |
@sonney2k | blackburn, for remote scheduling? | 22:17 |
blackburn | sonney2k: yes | 22:17 |
blackburn | sonney2k: like/dislike? | 22:18 |
blackburn | :) | 22:18 |
@sonney2k | like | 22:18 |
@sonney2k | but not sure | 22:18 |
blackburn | sonney2k: about? | 22:19 |
@sonney2k | where now() etc should be | 22:19 |
@sonney2k | I mean svm.train() will already train no? | 22:19 |
blackburn | sonney2k: svm.now().train()? | 22:19 |
blackburn | I am thinking about how to handle defaults here | 22:19 |
blackburn | by default it should be blocking (now()) | 22:19 |
@sonney2k | blackburn, and you want to have parameters for this time stuff | 22:19 |
blackburn | sonney2k: what do you mean? | 22:20 |
@sonney2k | e.g. memory limit | 22:20 |
@sonney2k | number of requested CPUs | 22:20 |
@sonney2k | when to run | 22:20 |
@sonney2k | use MPI | 22:20 |
@sonney2k | whatever | 22:20 |
blackburn | sonney2k: I am afraid I am overusing chains of method | 22:21 |
blackburn | s* | 22:21 |
@sonney2k | builderpattern without build lover | 22:21 |
blackburn | hah | 22:21 |
blackburn | sonney2k: it is probably d3.js that made me think about it | 22:24 |
@sonney2k | blackburn, but d3 is more crazy | 22:24 |
@sonney2k | kind of perceptron training in one line with data.enter().do_some_update() | 22:25 |
blackburn | sonney2k: yes but may be it appears to be better | 22:25 |
@sonney2k | it certainly is more compact | 22:26 |
@sonney2k | but also extremely hard to digest / read | 22:26 |
blackburn | sonney2k: tradeoff as always | 22:27 |
blackburn | sonney2k: recalling that model selection stuff | 22:27 |
blackburn | that builder-like thing is intended to fail during the run-time | 22:28 |
blackburn | and may be more error-prone | 22:28 |
blackburn | but much more readable and digestable | 22:28 |
blackburn | (0,1,0.1,R_EXP) | 22:28 |
@sonney2k | blackburn, why should at fail at run-time? | 22:30 |
blackburn | sonney2k: I am afraid there would be some things like overloaded methods | 22:31 |
blackburn | when you say use .from(10) on double parameter | 22:31 |
blackburn | I am even sure there would be some things that are compiling well but fail on runtime | 22:32 |
@sonney2k | ahh you are talking C++ | 22:32 |
blackburn | sonney2k: ah btw when do you think we can get to C++11? | 22:33 |
blackburn | sonney2k: have you learnt some C++11 btw? | 22:33 |
@sonney2k | blackburn, you want to totally drift away from C-style programming right? | 22:34 |
blackburn | sonney2k: depends what you mean by C-style | 22:35 |
@sonney2k | what we had | 22:35 |
@sonney2k | blackburn, regarding your api - in python modular that would not be an issue | 22:35 |
@sonney2k | it is just an exception at a different level | 22:36 |
blackburn | sonney2k: I am thinking about that C-style question | 22:36 |
blackburn | sonney2k: I don't have the general answer - but there are a few things I'd like to get rid of | 22:37 |
blackburn | like we all hate getters and setters and etc | 22:37 |
blackburn | sonney2k: I hope you do hate get/set too? | 22:37 |
@sonney2k | blackburn, I am pretty pragmatic | 22:38 |
@sonney2k | they do the job | 22:38 |
blackburn | sonney2k: they are disappearing in modern languages | 22:39 |
blackburn | like scala and etc | 22:39 |
@sonney2k | blackburn, replaced by what builder patterns? | 22:40 |
blackburn | sonney2k: no, why? | 22:40 |
@sonney2k | blackburn, no how do you set a variable then | 22:41 |
@sonney2k | in scala | 22:41 |
blackburn | sonney2k: obj.field = something | 22:41 |
@sonney2k | blackburn, public variables? | 22:41 |
@sonney2k | or what is the difference? | 22:41 |
blackburn | sonney2k: no this calls the getter | 22:41 |
blackburn | err | 22:41 |
blackburn | setter | 22:41 |
blackburn | autogenerated | 22:41 |
@sonney2k | wait so you still have a setter and getter? | 22:42 |
blackburn | sonney2k: not in your code | 22:42 |
@sonney2k | what setter/getter is then called? | 22:42 |
blackburn | sonney2k: autogenerated code | 22:42 |
blackburn | sonney2k: which you can overload if you really need some custom getter/setter | 22:43 |
@sonney2k | ok | 22:43 |
blackburn | sonney2k: if we were using only python | 22:44 |
blackburn | svm.parameter("C") = 3.0 | 22:44 |
blackburn | would be possible | 22:44 |
blackburn | or rather | 22:44 |
blackburn | svm["C"] | 22:44 |
blackburn | but we are tied to other langs | 22:44 |
@sonney2k | well svm.C would work if we don't want getters/setters at all | 22:45 |
blackburn | sonney2k: not in java/etc | 22:46 |
blackburn | haha I am still managing to produce at least one commit per day for 29 days | 22:47 |
blackburn | sonney2k: ah btw what about cmake | 22:58 |
blackburn | sonney2k: are we going to go for cmake some day? | 22:59 |
@sonney2k | as I said lot of work but I don't mind if someone does the job | 23:02 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 23:04 | |
@sonney2k | blackburn, please send the announcement | 23:28 |
blackburn | yes yes | 23:28 |
@sonney2k | would be bad if someone kills master with a patch now | 23:28 |
blackburn | ohh yes | 23:28 |
blackburn | well heiko reads mail usually | 23:29 |
@sonney2k | nite | 23:31 |
blackburn | alright | 23:32 |
blackburn | will send a mail in a few minutes | 23:32 |
--- Log closed Fri Apr 05 00:00:21 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!