--- Log opened Fri Jul 08 00:00:54 2011 | ||
-!- f-x` [~user@117.192.213.129] has quit [Remote host closed the connection] | 02:41 | |
-!- in3xes_ [~in3xes@210.212.58.111] has joined #shogun | 07:21 | |
-!- in3xes [~in3xes@210.212.58.229] has quit [Ping timeout: 252 seconds] | 07:24 | |
-!- sploving1 [~sploving@124.16.139.194] has joined #shogun | 08:30 | |
@sonney2k | sploving1, all good, i.e. making progress? | 08:47 |
---|---|---|
sploving1 | sonney2k, I am looking for generator.py | 08:48 |
sploving1 | looking generator.py | 08:48 |
@sonney2k | for or at? | 08:48 |
sploving1 | at. learning it | 08:49 |
sploving1 | os.path.insert? | 08:49 |
sploving1 | I did not see any docs about this | 08:49 |
sploving1 | sys.path.insert(0, '.') | 08:49 |
sploving1 | what does that mean? | 08:49 |
@sonney2k | sploving1, I mean you know what it is supposed to do. I don't think you can learn from that python example - I would suggest you just start fresh and write this in java | 08:50 |
@sonney2k | python doesn't really help you here... | 08:50 |
sploving1 | but I need understand what function that I should write | 08:51 |
@sonney2k | sploving1, but I told you and I know mikio too | 08:51 |
@sonney2k | again - just call the functions in each example | 08:52 |
@sonney2k | take what they return | 08:52 |
@sonney2k | and serialize this to disk | 08:52 |
@sonney2k | that's it | 08:52 |
@sonney2k | staring at python code won't help you with this | 08:53 |
sploving1 | why? understanding python is a quich way. | 08:53 |
sploving1 | I do not know how to serialize this to disk | 08:53 |
@sonney2k | http://www.google.com/search?q=java+serialize | 08:54 |
@sonney2k | and mikio is the java expert ask him | 08:55 |
sploving1 | oh | 08:56 |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 08:57 | |
@sonney2k | sploving1, better finish translating the examples first | 09:03 |
@sonney2k | I mean mostly only the more simple ones remain | 09:03 |
@sonney2k | like features_* | 09:03 |
@sonney2k | morning blackburn! | 09:03 |
@sonney2k | how's your bear? | 09:04 |
sploving1 | ofcourse I transfer all of them. | 09:04 |
sploving1 | they are easy. I like to do things changllge | 09:04 |
sploving1 | first | 09:04 |
blackburn | sonney2k: morgen! my bear is walking here somewhere | 09:04 |
@sonney2k | sploving1, yeah but it makes sense to finish one thing and then start the next | 09:04 |
@sonney2k | sploving1, in particular when a deadline is coming up and you don't have time to do more experimental stuff | 09:05 |
sploving1 | before the deadline, I will finishe them all. testsuite, docs and examples | 09:06 |
@sonney2k | sploving1, OK then just go ahead... and don't forget to ask if you get stuck (to not loose time...) | 09:07 |
sploving1 | oh | 09:07 |
sploving1 | sonney2k, I found there are *0.txt, *1.txt in testsuite tests. why 0 and 1? | 09:13 |
sploving1 | two test with the same mane | 09:13 |
sploving1 | I know. there are two para. | 09:13 |
sploving1 | parameters | 09:13 |
@sonney2k | sploving1, there are a number of parameter settings defined in the examples | 09:13 |
@sonney2k | functions are called for each setting | 09:13 |
@sonney2k | so we just index them with 0,1,2... | 09:14 |
sploving1 | oh. I know | 09:14 |
sploving1 | sonne2y, in java, also need set 2 parameter settings? | 09:34 |
@sonney2k | sploving1, as many as are specified in the example | 09:39 |
sploving1 | sonney2k, the I need change every example likt pytho using string[] parameter_list | 09:47 |
sploving1 | in the classifier_averaged_perceptron_modular.py example, why parameter_list = [[traindat,testdat,label_traindat,1.,1000,1],[traindat,testdat,label_traindat,1.,1000,1]] | 09:48 |
sploving1 | the two para is the same? | 09:48 |
@sonney2k | sploving1, you need to change every example anyways- turn them into functions | 09:48 |
@sonney2k | sploving1, yes that is stupid | 09:48 |
@sonney2k | a mistake I would say | 09:48 |
@sonney2k | please lower the first 1000 to e.g. 10 | 09:49 |
sploving1 | oh. | 09:49 |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 09:52 | |
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun | 09:55 | |
@sonney2k | heiko! | 09:56 |
@sonney2k | good morning | 09:56 |
heiko | hi sonney2k | 09:56 |
bettyboo | moin | 09:57 |
heiko | had a good sleep? :) | 09:57 |
@sonney2k | I might have found the correct way of saying friend to templates | 09:57 |
bettyboo | :) | 09:57 |
@sonney2k | heiko, didn't you know - I don't sleep | 09:57 |
heiko | yes? ok would be nice, I had a talk about this and some more thoughts, and I came to the conclusion that this does not work the way I want it | 09:57 |
heiko | oh well :) | 09:57 |
heiko | hehe, | 09:57 |
@sonney2k | heiko, it is committed already | 09:58 |
heiko | 12 hours of hacking followed by 12 hours of baby sitting :) | 09:58 |
heiko | ok I will have a look | 09:58 |
@sonney2k | if it works it was just a syntax thing | 09:58 |
sploving1 | sonney2k, you do not sleep???? | 09:58 |
@sonney2k | sploving1, get kids and you know what I mean | 09:59 |
sploving1 | I thought you are in regular work now | 09:59 |
sploving1 | great father! | 09:59 |
@sonney2k | yes once kids are there - they stay around :D | 10:00 |
bettyboo | :) sonney2k | 10:00 |
heiko | sonney2k, what did you change? | 10:00 |
@sonney2k | heiko https://github.com/shogun-toolbox/shogun/commit/37fcfdc4c5a3dcea44dbbc6d35c578f3866a2e92 | 10:01 |
heiko | so it is just the other template argument? | 10:03 |
heiko | weird | 10:03 |
@sonney2k | heiko, you can have one-to-one, one-to-many and many-to-many friend relation ships with templates | 10:04 |
* sonney2k who would have thought that | 10:05 | |
heiko | ok alright, this is really new to me :) | 10:06 |
heiko | thought you had to give the concrete class you want to be friend with | 10:06 |
heiko | . | 10:07 |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 10:08 | |
sploving1 | sonney2k, http://pastebin.com/BEtSdTTf | 10:09 |
sploving1 | I changed one example, and need your help to check if it is the correct what we expected | 10:09 |
heiko | could someone fix this: | 10:13 |
heiko | line 179 of Mathematics.cpp | 10:13 |
heiko | replace | 10:13 |
heiko | SGMatrix<float64_t> table(NULL,2,3); | 10:13 |
heiko | by | 10:13 |
heiko | SGMatrix<float64_t> table((float64_t*)NULL,2,3); | 10:13 |
heiko | I have other stuff in my master branch currently :( | 10:13 |
heiko | ambigious call compile error | 10:13 |
CIA-32 | shogun: Sergey Lisitsyn master * r57f2cfa / src/libshogun/lib/Mathematics.cpp : Fixed ambiguous ambiguity - http://bit.ly/nBFIwx | 10:23 |
blackburn | heiko: here you go ^) | 10:23 |
heiko | thanks :) | 10:24 |
bettyboo | ha ha!?! | 10:24 |
sploving1 | hello blackburn, in python example, we could return perceptron, out_labels. then in java, how do we return two object? | 10:25 |
blackburn | return collection containing two objects if it is necessary | 10:25 |
sploving1 | which collection can store different type of objects? | 10:29 |
sploving1 | ArrayList is good | 10:30 |
blackburn | yes, it is ok | 10:30 |
blackburn | all collections are generic | 10:31 |
sploving1 | blackburn, what about parameters? for instance, in python, parameter_list = [[traindat,testdat,label_traindat,1.,1000,1],[traindat,testdat,label_traindat,1.,1000,1]] | 10:33 |
sploving1 | classifier_averaged_perceptron_modular(*parameter_list[0]) | 10:33 |
sploving1 | if I want to implement similar parameter_list, then use ArrayList also? | 10:33 |
heiko | blackburn, sonney2k, dynamic_casts allowed in shogun? | 10:33 |
blackburn | sploving1: yes, or collection of array lists :) | 10:33 |
blackburn | heiko: where do you need it?? :) | 10:34 |
heiko | in my new DynamicObjectArray | 10:35 |
heiko | to ensure that generic type T is of type CSGObject | 10:35 |
blackburn | I don't know.. anyway you can change it later if it is bad | 10:36 |
heiko | yes | 10:36 |
heiko | will just use | 10:36 |
sploving1 | blackburn, ArrayList parameterOne = new ArrayList(); parameterOne.add(new Double(1.0)); parameterOne.add(new Integer(1000)); | 10:53 |
sploving1 | then how to get 1.0 and 1000 from the list? | 10:54 |
blackburn | .get(0), get(1), ... | 10:54 |
sploving1 | need type converstion. int a = parapameterOne.get(0), will be OK? blackburn | 10:55 |
blackburn | int a = ((Integer)parameterOne.get(1)).intValue(); | 10:55 |
blackburn | something like that | 10:55 |
sploving1 | so complex | 10:55 |
blackburn | just conversion from Integer to int | 10:56 |
blackburn | and typecast - get() returns Object - it is generic | 10:57 |
sploving1 | oh | 10:57 |
-!- VojtechFranc [~quassel@gw-101.scnet.cz] has joined #shogun | 11:07 | |
-!- sploving1 [~sploving@124.16.139.194] has left #shogun [] | 11:28 | |
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 12:01 | |
petiera | hallo! could you help me get started with this toolbox? If I already have similarity-matrices and would like to use a CombinedKernel, is it enough to just make CustomKernels using the matrices and then append them to the CombinedKernel? Or do I have to have Features as well? | 12:20 |
blackburn | sonney2k: ^ | 12:20 |
-!- in3xes1 [~in3xes@180.149.49.227] has joined #shogun | 12:30 | |
-!- in3xes_ [~in3xes@210.212.58.111] has quit [Ping timeout: 252 seconds] | 12:32 | |
-!- f-x [~user@117.192.202.110] has joined #shogun | 12:35 | |
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has joined #shogun | 12:40 | |
-!- VojtechFranc [~quassel@gw-101.scnet.cz] has quit [Remote host closed the connection] | 12:46 | |
@sonney2k | blackburn, only the tough questions for me? | 13:05 |
@sonney2k | petiera, I am not sure - in the worst case create some DummyFeatures within a CombinedFeatures object | 13:06 |
@sonney2k | petiera, if get_kernel_matrix works w/o then yes you are fine :) | 13:06 |
blackburn | sonney2k: no, the questions I can't answer :) | 13:07 |
@sonney2k | sploving1, is it possible to use non-ascii too? I mean not just strings but tuples utilizing the actual types? | 13:07 |
sploving1 | I am tring ArrayList | 13:08 |
sploving1 | sonney2k, which can contain various types | 13:08 |
@sonney2k | otherwise you have to the conversion in the example (I am not a big fan of this...) | 13:09 |
@sonney2k | sploving1, sounds good | 13:09 |
@sonney2k | hmmhh heiko is gone already? | 13:10 |
@sonney2k | ! | 13:10 |
blackburn | sonney2k: which conversion? | 13:10 |
@sonney2k | "1" -> 1 | 13:10 |
petiera | sonney2k: ok, thanks, I'll try that | 13:11 |
blackburn | sonney2k: you make me scared, string 'typization' is scaring | 13:11 |
@sonney2k | blackburn, yeah but that is why I don't like sploving1 suggestions that much ( http://pastebin.com/BEtSdTTf ) | 13:14 |
@sonney2k | petiera, and let us know what the outcome is please! | 13:15 |
blackburn | ahh | 13:15 |
@sonney2k | petiera, I remember doing some hacks that features are not required but I cannot remember and we don't have an example for combined custom kernels yet... | 13:15 |
sploving1 | sonney2k, http://pastebin.com/43zkc1Ts | 13:18 |
sploving1 | take a look at the new version | 13:18 |
@sonney2k | blackburn, ^ can one write things more compact? | 13:19 |
@sonney2k | the parameter_list is quite longish now | 13:19 |
blackburn | sonney2k: I don't think so | 13:19 |
blackburn | sonney2k: Collections.asList(new Double(10), new Integer(1000)); -- better? | 13:20 |
petiera | sonney2k: yeah, I tried to look for examples but didn't find anything similiar there. I tried it now with two random square matrices and I do get a matrix out of the CombinedKernel's get_kernel_matrix, however the weights are both 1.0 | 13:21 |
@sonney2k | petiera, both? | 13:22 |
petiera | of course the input matrices are just random, but I'd still expect some deviations | 13:22 |
@sonney2k | what do you mean | 13:22 |
petiera | yes, both | 13:22 |
@sonney2k | petiera, what does both mean | 13:22 |
@sonney2k | petiera, put your code to pastebin or so | 13:22 |
petiera | sonney2k: aa, sorry :) it means that when I give two CustomKernels to the CombinedKernel and train it with SVM, the subkernel weights are both 1.0 | 13:23 |
@sonney2k | petiera, sure | 13:23 |
@sonney2k | the svm won't change them | 13:23 |
sploving1 | blackburn, you mean : parameterOne = Collections.asList(new Double(10), new Integer(1000)); | 13:24 |
petiera | http://pastebin.com/GwAgXv9j | 13:24 |
blackburn | well may be just simply | 13:24 |
blackburn | parameter_list.add(Collections.asList(...)); | 13:24 |
blackburn | it could be shorter and should work | 13:25 |
blackburn | will test it in a min | 13:25 |
@sonney2k | that is slightly better | 13:25 |
blackburn | we can't do non-homogenous collection of simple types | 13:25 |
blackburn | so that new Double is required unfortunately | 13:26 |
petiera | sonney2k: so does it mean that the CombinedKernel isn't optimized? | 13:26 |
blackburn | sploving1: you haven't initialized your parameter list | 13:27 |
blackburn | use not public ArrayList parameter_list; | 13:27 |
sploving1 | fixed parameter_list = new ArrayList(); | 13:27 |
blackburn | but public Collection parameter_list = new ArrayList(2); | 13:27 |
@sonney2k | petiera, yes of course not | 13:28 |
@sonney2k | why should it be changed? | 13:28 |
@sonney2k | petiera, it is just an SVM with two kernels | 13:28 |
@sonney2k | perhaps you want to do MKL? | 13:28 |
blackburn | sploving1: ah yes you shouldn't use Collection, it has no get | 13:30 |
petiera | sonney2k: Yes, that would be my intention | 13:30 |
sploving1 | blackburn, then use what? | 13:31 |
blackburn | sploving1: | 13:31 |
blackburn | ArrayList pars = new ArrayList(2); | 13:31 |
blackburn | pars.add(Arrays.asList(new Double(1.0), new Integer(1000))); | 13:31 |
blackburn | sth like that | 13:31 |
petiera | sonney2k: ahh, so probably MKLClassifier? :) | 13:31 |
bettyboo | hoho | 13:31 |
@sonney2k | yep | 13:31 |
petiera | hehe. thanks for help! As I said, I'm just getting to know this toolbox :) | 13:33 |
sploving1 | blackburn, then how to cast type pars.get(0) | 13:34 |
blackburn | sploving1: just List | 13:34 |
sploving1 | then how to get List 's first elment? | 13:35 |
sploving1 | blackburn, List x= .., x[0]?? | 13:35 |
blackburn | ((List)parameters.get(0)).get(0) | 13:36 |
sploving1 | sonney2k, blackburn, I found org.shogun.* also hava List | 13:37 |
sploving1 | class org.shogun.List | 13:37 |
sploving1 | double learn_rate = ((Double)para.get(0)).doubleValue() | 13:37 |
sploving1 | error | 13:37 |
@sonney2k | sploving1, but it is typed | 13:37 |
@sonney2k | better not use shogun lists | 13:37 |
@sonney2k | for this purpose | 13:37 |
sploving1 | sonney2k, I do not mean to use shogun list, but its used by the system | 13:39 |
blackburn | import java.util.List | 13:39 |
sploving1 | worked! http://pastebin.com/cMTZhQ7K | 13:41 |
sploving1 | sonney2k, do you think this example could be used by generator.py? | 13:42 |
sploving1 | generator.java(maybe) | 13:42 |
sploving1 | the example has no return type in main function!!!! | 13:42 |
@sonney2k | sploving1, do you need to do new ArrayList(2); and then add? | 13:43 |
blackburn | sonney2k: he could init it just where it is declared | 13:43 |
@sonney2k | sploving1, it looks good to me otherwise - I would group the shogun imports and java imports to beautify things a bit | 13:44 |
@sonney2k | blackburn, I don't understand? | 13:44 |
blackburn | sonney2k: line 12 | 13:44 |
sploving1 | public ArrayList parameter_list = new ArrayList(2); | 13:44 |
sploving1 | you mean this? | 13:44 |
blackburn | yes | 13:44 |
@sonney2k | ahh ok yes | 13:44 |
sploving1 | sounds good | 13:44 |
sploving1 | how to group them sonney2k. | 13:45 |
@sonney2k | sploving1, I mean all the imports java* behind the shogun ones | 13:46 |
sploving1 | I know. java.util first | 13:46 |
sploving1 | oh | 13:46 |
sploving1 | The last thing I worried is that this example could be called by generator.java and tester.java | 13:47 |
sploving1 | if can, I will then change all the other examples | 13:47 |
sploving1 | so I would first write the genetrator.java and tester.java to test this example | 13:48 |
@sonney2k | it is still nicely readable - which is good :) | 13:54 |
-!- heiko [~heiko@main.uni-duisburg.de] has joined #shogun | 13:59 | |
@sonney2k | heiko, did the template thing work? | 14:01 |
heiko | sonney2k, hi, yes compiles | 14:01 |
@sonney2k | great | 14:01 |
heiko | I step wise replace stuff | 14:01 |
heiko | but I got a nasty memory bug currently | 14:01 |
@sonney2k | heiko, I wasn't sure what you meant me to do with your last pull request | 14:01 |
heiko | It was just to show the code, just forget it | 14:02 |
@sonney2k | ok closing then | 14:02 |
heiko | When the DynArray->DynamicObjectArray replacement is done for all places, python will hopefully run | 14:02 |
petiera | sonney2k: now when I call train() on a MKLClassification-object with a CombinedKernel, I get "Interleaved MKL optimization is currently only supported with SVMlight | 14:02 |
heiko | think I already did? | 14:02 |
@sonney2k | yes | 14:02 |
@sonney2k | you did :) | 14:02 |
@sonney2k | petiera, which says it all... so disable it | 14:03 |
@sonney2k | or enable SVMlight | 14:03 |
petiera | sonney2k: I'm using python, so should I recompile the C++? | 14:03 |
petiera | I mean if I want to enable SVMlight | 14:04 |
@sonney2k | if you need it excessively yes ./configure --interfaces=libshogun,python_modular | 14:04 |
@sonney2k | yes | 14:04 |
petiera | ok | 14:04 |
petiera | thanks again for helping, and have a nice weekend! | 14:14 |
-!- petiera [~ohyvarin@kosh.org.aalto.fi] has left #shogun [] | 14:14 | |
sploving1 | sonney2k, are you around? | 15:05 |
@sonney2k | yes | 15:05 |
sploving1 | how to edit the doc? | 15:05 |
sploving1 | I would add doc first when waiting for the email | 15:05 |
@sonney2k | sploving1, go to doc/pages directory | 15:06 |
sploving1 | I mean http://www.shogun-toolbox.org/doc/installation.html this page | 15:06 |
@sonney2k | then edit Installation.mainpage | 15:06 |
sploving1 | sonne2k, I thought online editing | 15:07 |
@sonney2k | ? | 15:07 |
@sonney2k | no it is all doxygen | 15:07 |
@sonney2k | just run make doc in doc/ to get the updated pages | 15:08 |
@sonney2k | blackburn, ahh I just realize that my nice automagic make release script won't work any longer | 15:08 |
@sonney2k | that all is still svn based | 15:09 |
blackburn | sonney2k: which oe? | 15:09 |
blackburn | one* | 15:09 |
blackburn | make release? | 15:09 |
blackburn | what is it? | 15:09 |
@sonney2k | that will take some time to get this reliably working with git too | 15:09 |
@sonney2k | blackburn, it did run tests | 15:09 |
@sonney2k | modify the version and tag the release | 15:09 |
@sonney2k | then upload everything gpg signed | 15:10 |
blackburn | oh | 15:10 |
blackburn | I see | 15:10 |
@sonney2k | including docs and webpage changes | 15:10 |
blackburn | sonney2k: I guess sploving1 will write another one JUnit :D | 15:19 |
sploving1 | blackburn, JUnit? never heard | 15:20 |
@sonney2k | blackburn, please not | 15:20 |
blackburn | sploving1: framework for testing :) | 15:20 |
@sonney2k | it is totally sufficient to have regression tests like this - if it works | 15:20 |
@sonney2k | in the worst case we will serialize shogun objects and compare if they are different | 15:21 |
blackburn | sonney2k: I mean that what he will write - is kind of 'truncated' JUnit :) | 15:21 |
@sonney2k | blackburn, I hope it is very lightweigth, i.e. just ~100 lines java | 15:21 |
@sonney2k | not a JUnit monster | 15:21 |
blackburn | sonney2k: non-trivial thing :) | 15:22 |
bettyboo | :*) | 15:22 |
@sonney2k | blackburn, just looking at their github site - JUnit has 134 issues... more than we can ever have | 15:23 |
@sonney2k | ;-) | 15:23 |
blackburn | sonney2k: sure, it is half-joke but it is something we do | 15:23 |
blackburn | btw we should have some libshogun tester | 15:24 |
sploving1 | sonney2k, pulled the Install.Main page | 15:24 |
-!- heiko [~heiko@main.uni-duisburg.de] has quit [Ping timeout: 258 seconds] | 15:25 | |
sploving1 | sonney2k, other doc page that I need to update | 15:29 |
@sonney2k | Tutorial | 15:29 |
-!- blackburn [~blackburn@188.122.238.99] has quit [Quit: Leaving.] | 15:38 | |
@sonney2k | ahh sploving the INSTALL file in /src/ | 15:41 |
sploving1 | oh | 15:51 |
sploving1 | sonney2k, Turorial use classifier_libsvm_modular? | 15:54 |
@sonney2k | sploving1, yes | 15:54 |
@sonney2k | all languages use this for illustration | 15:54 |
sploving1 | oh | 15:54 |
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has quit [Read error: Connection reset by peer] | 16:07 | |
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has joined #shogun | 16:25 | |
sploving1 | my machine crash when open for more than 4 hours | 16:26 |
@sonney2k | sploving1, great! | 16:26 |
@sonney2k | looks like everyones computer is dying here once per GSoC | 16:26 |
* sonney2k just bought a new harddrive | 16:27 | |
@sonney2k | IIRC heiko had to reinstall his computer too | 16:27 |
sploving1 | oh. my notebook just used for one year | 16:28 |
sploving1 | it is still new | 16:28 |
-!- phrhbk [~asdfasdf@host-57-20.warwick.ac.uk] has joined #shogun | 16:32 | |
@sonney2k | sploving1, then you still have warrenty! | 16:45 |
sploving1 | sonney2k, ?? warrenty? | 16:46 |
@sonney2k | warranty! | 16:46 |
sploving1 | I forgot where I put it. | 16:48 |
sploving1 | sonney2k, For octave: | 16:49 |
sploving1 | \verbinclude classifier_libsvm_minimal_modular.m | 16:49 |
sploving1 | then how to add the example? | 16:49 |
sploving1 | java example | 16:49 |
sploving1 | I think it is in the example dir. so just edit it like octave | 16:50 |
sploving1 | sonney2k, pushed the doc | 17:12 |
@sonney2k | sploving1, exactly | 17:22 |
sploving1 | sonney2k, have you seen the email that Miko sent me? | 17:30 |
@sonney2k | no | 17:31 |
-!- in3xes1 is now known as in3xes | 17:42 | |
-!- sploving1 [~sploving@2001:cc0:2020:2022:5eff:35ff:fe04:f091] has left #shogun [] | 17:44 | |
-!- Guest74997 [~f-x@213.155.190.134] has quit [Read error: Connection reset by peer] | 18:26 | |
-!- f-x [~user@117.192.202.110] has quit [Remote host closed the connection] | 18:52 | |
-!- f-x [~user@117.192.202.110] has joined #shogun | 19:44 | |
-!- blackburn [~blackburn@188.122.238.99] has joined #shogun | 19:53 | |
blackburn | sonney2k: have you read mikio's answer to baozeng? | 19:54 |
-!- f-x [~user@117.192.202.110] has quit [Read error: Connection reset by peer] | 20:34 | |
blackburn | http://www.getacoder.com/projects/solve%20p%20vs%20np_132036.html | 20:37 |
in3xes | hehe | 21:14 |
in3xes | funny | 21:15 |
@sonney2k | blackburn, yes | 21:26 |
@sonney2k | I am not entirely enlightened though | 21:26 |
blackburn | sonney2k: very nice idea - to use reflection and load class | 21:27 |
blackburn | sonney2k: btw tell baozeng about codestyle if I forgot - he does it non java way :) | 21:28 |
blackburn | camelCase! | 21:29 |
blackburn | it is much better - I guess running java ~100 times is worse than load 100 classes | 21:31 |
blackburn | hmm | 21:35 |
-!- f-x [~user@117.192.202.110] has joined #shogun | 21:35 | |
blackburn | no, no difference | 21:35 |
blackburn | just tests | 21:35 |
-!- in3xes is now known as gayam | 21:44 | |
-!- gayam is now known as in3xes | 21:44 | |
CIA-32 | shogun: Baozeng Ding master * r1dac476 / doc/pages/Installation.mainpage : add java support in Installation.mainpage - http://bit.ly/oLA0R7 | 21:44 |
CIA-32 | shogun: Baozeng Ding master * r607aae4 / doc/pages/ModularTutorial.mainpage : add ModularTutorial mainpage for java - http://bit.ly/r80wrD | 21:44 |
CIA-32 | shogun: Baozeng Ding master * r9fbef9c / src/INSTALL : add java support in INSTALL file - http://bit.ly/pqGWhR | 21:44 |
CIA-32 | shogun: Soeren Sonnenburg master * r1119fc2 / (3 files in 2 dirs): | 21:44 |
CIA-32 | shogun: Merge pull request #183 from sploving/master | 21:44 |
CIA-32 | shogun: add ModularTutorial mainpage - http://bit.ly/qMMkRT | 21:44 |
@sonney2k | blackburn, I actually don't care so much... in python we just import all the *.py files | 21:45 |
@sonney2k | and then run the tests | 21:45 |
@sonney2k | blackburn, what I don't know is if one can serialize objects in java without deriving from some base class | 21:46 |
@sonney2k | I wouldn't really like making the examples even more difficult | 21:46 |
blackburn | Serializable is interface IIRC | 21:47 |
@sonney2k | blackburn, ok so it is not that bad one just writes implements Serializable right? | 21:48 |
blackburn | sonney2k: base class Test implements Serializable, that's all | 21:48 |
blackburn | why it could be bad? it is ok for me ;) | 21:48 |
@sonney2k | blackburn, wait does mean the generator can do this or has this to be done in the examples? | 21:49 |
blackburn | sonney2k: in examples we have to have some base class where IO operations will be implemented | 21:49 |
blackburn | sonney2k: when in generator we could just readObject and writeObject | 21:50 |
blackburn | without any care | 21:50 |
@sonney2k | blackburn, yeah but that makes examples harder to read | 21:50 |
blackburn | why?? examples will stay simple | 21:50 |
@sonney2k | didn't you say we need to write some io operations? | 21:51 |
blackburn | sonney2k: only in base class | 21:51 |
blackburn | or even without it | 21:51 |
@sonney2k | so you would derive each example from that base class | 21:51 |
blackburn | yes | 21:51 |
@sonney2k | I would love to not have to do that though | 21:51 |
blackburn | why? | 21:52 |
@sonney2k | would be great if examples could stand alone | 21:52 |
blackburn | what do you mean saying alone? | 21:52 |
@sonney2k | blackburn, no dependencies | 21:53 |
@sonney2k | and if then only for loading data | 21:53 |
blackburn | all examples could implement that run() and have main for case it is run with java -jar someExample | 21:53 |
blackburn | dependencies from? | 21:53 |
blackburn | Test base class is much simpler for generator | 21:56 |
blackburn | it just load it and doing all the things that lovers do | 21:56 |
blackburn | well really we can implement run everywhere | 21:58 |
blackburn | almost the same | 21:58 |
* blackburn is implementing heap for shogun | 22:04 | |
f-x | sonney2k: are you there? | 23:27 |
--- Log closed Sat Jul 09 00:00:57 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!