--- Log opened Wed Aug 17 00:00:59 2011 | ||
@sonney2k | blackburn, arg my machine rebooted spontaneously again | 00:02 |
---|---|---|
@sonney2k | and I did a memory test for the last few hours to check - no errors | 00:03 |
@sonney2k | increased fan speed | 00:03 |
@sonney2k | cpu temp down to ~25 | 00:03 |
@sonney2k | still it crashes | 00:03 |
blackburn | strange | 00:03 |
blackburn | HR! | 00:03 |
blackburn | works for parallel | 00:03 |
blackburn | but not for SGIO?! | 00:04 |
@sonney2k | running linux 3.0.2 now | 00:04 |
@sonney2k | lets see if this changes things | 00:04 |
blackburn | WTF is wrong with SGIO | 00:08 |
@sonney2k | blackburn, could it be that you have to %extend before the class is %included? | 00:09 |
@sonney2k | anyway I am too tired for anything now | 00:10 |
blackburn | no but I have an idea it is included after extended | 00:10 |
@sonney2k | cu | 00:10 |
blackburn | see you | 00:10 |
serialhex | hey blackburn whats up? | 00:19 |
blackburn | serialhex: hi having a luv time with SGIO things, how do you do? | 00:19 |
serialhex | i'm doing well.. the pc i set upto run debian so i could do SG work on it died :( | 00:20 |
serialhex | more accurately: the 4G hdd died, but the rest of the hardware works fine afaik | 00:20 |
serialhex | SGIO == shogun input/output?? | 00:21 |
blackburn | yeah | 00:21 |
serialhex | ahh, cool cool | 00:22 |
blackburn | for some strange reason it doesn't work with java externalization stuff | 00:22 |
serialhex | hmm... sounds fun! :P | 00:22 |
blackburn | serialhex: I wanted to ask you to read description I wrote if you have spare time | 00:22 |
blackburn | for non-english english :D | 00:23 |
serialhex | for the docs & explanations of things in shogun right? | 00:24 |
blackburn | yeah | 00:24 |
serialhex | i might not always have time to come in here and have ya ask that, so just feel free to e-mail them to me and i'll read them as soon as i can & let ya know | 00:24 |
blackburn | I am not sure it is parseable for other people haha | 00:25 |
serialhex | i have libre office if that makes any difference for ya | 00:25 |
blackburn | so okay | 00:25 |
blackburn | preprocessor/MultidimensionalScaling.h | 00:25 |
blackburn | preprocessor/Isomap.h | 00:25 |
blackburn | preprocessor/LocalllyLinearEmbedding.h | 00:26 |
blackburn | preprocessor/LocalTangentSpaceAlignment.h | 00:26 |
blackburn | preprocessor/HessianLocallyLinearEmbedding.h | 00:26 |
blackburn | preprocessor/LaplacianEigenmaps.h | 00:26 |
blackburn | there are a short description starting with @brief | 00:26 |
blackburn | I would appreciate if you let me know of mistakes I did here | 00:27 |
serialhex | what? it's actually in the header files? | 00:27 |
blackburn | yeah | 00:27 |
serialhex | ooh!!! ur gonna vex me!!! :P | 00:27 |
blackburn | we have much doc in .h :) | 00:27 |
blackburn | well we have *all* doc in .h files | 00:27 |
serialhex | no problem, it's in shogun upstream right? | 00:28 |
blackburn | yeah | 00:28 |
serialhex | alright then, as you do them & push the docs just send me an e-mail... you got my addy? | 00:28 |
blackburn | it is already there | 00:28 |
blackburn | I can sent you cut-paste of descriptions by email | 00:29 |
blackburn | if it is more convenient for you | 00:29 |
blackburn | sent | 00:29 |
blackburn | to you | 00:29 |
blackburn | sent | 00:29 |
blackburn | send | 00:29 |
blackburn | :D | 00:29 |
serialhex | either way is cool... oryou can post a link to the commit on github so i can read it there... | 00:30 |
serialhex | that'd probably be easiest for both of us! | 00:30 |
blackburn | https://github.com/shogun-toolbox/shogun/commit/9d2dd548c70bff45d8ac5a1aa00556ca425ef365 | 00:31 |
blackburn | serialhex: the main question heh | 00:32 |
blackburn | if I say description is given in some paper | 00:32 |
blackburn | or at some paper? | 00:32 |
blackburn | in/at/..? | 00:32 |
serialhex | in some paper | 00:32 |
blackburn | at page? | 00:33 |
serialhex | in some *thing*; at some *place* would be a general rule | 00:33 |
blackburn | hmm nice | 00:33 |
blackburn | thanks | 00:33 |
serialhex | i'm in a box, at a bar :P | 00:33 |
blackburn | is "at page 261" ok to say? | 00:34 |
serialhex | i'll take a look at these and let ya know, do you want me to add comments directly into the source on github or would you prefer if i e-mailed them to ya? | 00:34 |
serialhex | no, "on page 261" would be better | 00:35 |
blackburn | on! | 00:35 |
blackburn | ohh | 00:35 |
blackburn | serialhex: comment on github would be more convenient for you | 00:35 |
serialhex | yeah :) | 00:35 |
serialhex | yes, but it will also be there forever in the shogun commit data (i think) | 00:35 |
serialhex | (at least the github commit stuff) | 00:36 |
blackburn | you could say it right to me here | 00:36 |
blackburn | yeah shameful history heh | 00:36 |
serialhex | well i'm going to eat dinner in a few... how late is it there? or more importantly: how late are you staying up?? | 00:36 |
blackburn | 2-36 | 00:37 |
blackburn | I don't know | 00:37 |
serialhex | yeah, that's the only reason i wouldn't want to just post a bunch of grmatical edits to github... though having you here while i edit it would help us both i think | 00:37 |
blackburn | if you will be back in 20-30 min we could make it today | 00:37 |
serialhex | ok, will do! | 00:37 |
serialhex | bbl!! | 00:37 |
blackburn | hehe | 00:37 |
blackburn | again wrote if + future | 00:37 |
blackburn | WTF IS WRONG WITH SGIO :( | 00:38 |
serialhex | ok, i'm back | 00:55 |
serialhex | lemme see whats up with these docs blackburn | 00:56 |
blackburn | okay let's start with hmm | 00:56 |
blackburn | serialhex: https://github.com/shogun-toolbox/shogun/blob/9d2dd548c70bff45d8ac5a1aa00556ca425ef365/src/shogun/preprocessor/MultidimensionalScaling.h | 00:57 |
blackburn | lines 25-53 | 00:57 |
serialhex | need an 'is': /** @brief the class Multidimensionalscaling is used to perform | 00:58 |
blackburn | fixed | 00:58 |
blackburn | hey thanks in advance you help me so much | 00:58 |
serialhex | at -> on * Description of classical embedding is given on p.261 (Section 12.1) of | 00:58 |
serialhex | added 'the': * In this preprocessor the LAPACK routine DSYEVR is used for | 00:59 |
serialhex | added 'an' * solving an eigenproblem. If ARPACK library is available, | 01:00 |
serialhex | i'll private gist the whole thing when i'm done too... | 01:00 |
blackburn | hm | 01:00 |
blackburn | okay | 01:00 |
serialhex | ok, the next para needs to be revised a bit, hd on | 01:02 |
CIA-87 | shogun: Sergey Lisitsyn master * r6e008b4 / src/interfaces/modular/modshogun_ignores.i : Temporary? hack for JAVA externalization - http://bit.ly/rnTRvL | 01:02 |
blackburn | why I wrote JAVA hahah | 01:04 |
serialhex | ok, quick Q: on this line " * might be inconsistent (essentially, according to zero or" & the next, it would be 'having a zero or negative eigenvalue'? is thar about it? | 01:05 |
serialhex | i just want to make sure my understanding is correct | 01:06 |
blackburn | hmm | 01:06 |
blackburn | okay feature here is eigenvector | 01:06 |
blackburn | so I tried to say that features=eigenvectors according to non-pos values are inconsistent or so | 01:07 |
serialhex | ok, so features=eigenvectors and non-pos values are inconsistent | 01:08 |
serialhex | is that what you mean when you say 'according to'? | 01:08 |
blackburn | yeah features according to neg values | 01:08 |
blackburn | I don't sure it is right word | 01:08 |
-!- f-x [~user@117.192.215.151] has quit [Remote host closed the connection] | 01:09 | |
serialhex | 'i'm not sure if it is the right word' :P | 01:09 |
blackburn | yeah I noticed that just after wrote | 01:09 |
serialhex | ok, here it is: https://gist.github.com/16e77bf6414e569e7214 | 01:12 |
blackburn | thanks a lot! | 01:12 |
serialhex | hmm... i wonder if there is a way to diff gists...? | 01:13 |
blackburn | I just copied | 01:13 |
serialhex | np... i just figured so you would know what i changed so you could learn better, ask me Q's qhy i changed X or whatever | 01:14 |
blackburn | okay ready for another one? | 01:14 |
blackburn | https://github.com/shogun-toolbox/shogun/blob/9d2dd548c70bff45d8ac5a1aa00556ca425ef365/src/shogun/preprocessor/LocallyLinearEmbedding.h | 01:16 |
serialhex | shoot | 01:16 |
blackburn | good thing is I see many mistakes here :D | 01:16 |
blackburn | but not all of them could be fixed by me | 01:16 |
serialhex | Q: Available from, 290(5500), 2323-2326. how do you know where to get the article from these numbers??? | 01:21 |
serialhex | (idk if it's wrong, i'm just curious) | 01:21 |
blackburn | well it is from mendeley | 01:22 |
blackburn | copied as is | 01:22 |
blackburn | hmm but yes I could put a different link here | 01:23 |
blackburn | Saul, L. K., Ave, P., Park, F., & Roweis, S. T. (2001). An Introduction to Locally Linear Embedding. Available from, 290(5500), 2323-2326. Citeseer. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.123.7319&rep=rep1&type=pdf | 01:23 |
blackburn | that's better | 01:23 |
serialhex | i'll put that in there for ya... | 01:23 |
blackburn | thanks | 01:24 |
serialhex | ooh, that last para is nice :P | 01:27 |
blackburn | did you mean crazy inparseable shit? | 01:28 |
serialhex | yeah kind of :) | 01:30 |
serialhex | https://gist.github.com/7a8015ea2d7c520aa880 make sure that last para is right | 01:30 |
blackburn | almost right but ARPACK is being used with regularization | 01:31 |
blackburn | not for | 01:31 |
serialhex | ahh, ok | 01:31 |
blackburn | hmm how to say dat riht :D | 01:31 |
blackburn | with small regularization of weight matrix | 01:31 |
blackburn | Due to computation speed, ARPACK is being used with small regularization of weight matrix and Cholesky factorization of it is used internally for Lanzcos iterations. If the results aren't reasonable LUP factorization could be used with posdef parameter set to false using set_posdef. | 01:32 |
blackburn | serialhex: ok? ^ | 01:33 |
serialhex | https://gist.github.com/7a8015ea2d7c520aa880 | 01:33 |
serialhex | yeah i think what i just sent is ~ the same :P | 01:33 |
serialhex | either will work | 01:33 |
blackburn | okay | 01:33 |
blackburn | thanks | 01:33 |
serialhex | next | 01:34 |
blackburn | small ones left | 01:35 |
blackburn | https://github.com/shogun-toolbox/shogun/blob/9d2dd548c70bff45d8ac5a1aa00556ca425ef365/src/shogun/preprocessor/LocalTangentSpaceAlignment.h | 01:35 |
serialhex | ok cool | 01:35 |
blackburn | see one mistake already heh | 01:36 |
serialhex | yeah, you've got a weird char here: * (2004) 313â338. :P | 01:37 |
blackburn | hmm | 01:38 |
blackburn | no idea what is it | 01:38 |
blackburn | will get better citation in a min | 01:39 |
serialhex | https://gist.github.com/ba1542ead92167f02797 not much but double check to see that i got it right | 01:39 |
blackburn | yeah all ok | 01:41 |
serialhex | in fact, for future ones you can probably private gist them to me and i'll fork & edit and e-mail ya back the links for ones later on | 01:43 |
blackburn | serialhex: https://github.com/shogun-toolbox/shogun/blob/9d2dd548c70bff45d8ac5a1aa00556ca425ef365/src/shogun/preprocessor/LaplacianEigenmaps.h short one too | 01:43 |
blackburn | that one and Isomap and over | 01:47 |
blackburn | ah nothing to check in isomap | 01:47 |
blackburn | last one | 01:47 |
blackburn | ^ | 01:47 |
serialhex | LAPACK DSYGVX... could they come up with a weirder name??? | 01:47 |
blackburn | Double SYmmetric Generalized VX haha | 01:49 |
blackburn | no idea what VX is | 01:49 |
blackburn | e.g. DSYEV stands for Double Symmetic Eigenvalues | 01:49 |
blackburn | DSYEVR same but RRR method | 01:49 |
serialhex | https://gist.github.com/50837ac8b233fa396be1 make sure this is right again... i changed some wording and may have mangled the meaning :P | 01:50 |
blackburn | Note that the algorithm is very sensitive to heat distribution coefficient | 01:51 |
blackburn | shouldn't it be | 01:51 |
blackburn | Note that the algorithm is very sensitive to the heat distribution coefficient | 01:51 |
blackburn | ? | 01:51 |
serialhex | yeah, derr! | 01:51 |
blackburn | same for the number of neighbor, right? | 01:51 |
blackburn | oh nonono | 01:52 |
blackburn | When no connectivity check is * provided the preprocessor will not produce reasonable embeddings if K value * makes a graph that is not connected. | 01:52 |
serialhex | https://gist.github.com/50837ac8b233fa396be1 | 01:52 |
serialhex | try that | 01:52 |
blackburn | I tried to say that there are no connectivity check so if K is not sufficient it will be fucked up | 01:53 |
serialhex | yeah, thats the gist of it... | 01:54 |
blackburn | hmm my bad understanding | 01:55 |
blackburn | ahh | 01:55 |
blackburn | serialhex: so saying when blablabla is not a condition? | 01:55 |
blackburn | I mean in this sentence it is stated that no check here, right? | 01:56 |
serialhex | no it's a condition... | 01:56 |
blackburn | it shouldn't be a condition | 01:56 |
blackburn | should be something like - there is no connectivity check - be careful :) | 01:57 |
serialhex | yeah, when there is no check then shit will be fucked up if the k value makes a graph that is not connected | 01:57 |
blackburn | so is this right now? | 01:58 |
serialhex | yes | 01:58 |
blackburn | okay | 01:58 |
serialhex | as long as we both understand that you meant that shit will be fucked up if you dont do the connectivity check :P | 01:59 |
blackburn | hmm | 02:00 |
blackburn | does it say that there is no option to *do* that check? | 02:00 |
blackburn | that's what confusing me | 02:00 |
blackburn | because when translating to my lang it looks like "if you do the check all will be good, if not - it could fuck up embedding" | 02:01 |
blackburn | ah what a crazy discussion :D | 02:01 |
serialhex | yeah thats about it :) | 02:01 |
serialhex | and yes it is crazy, but we're getting there | 02:02 |
blackburn | so finally | 02:02 |
blackburn | is no ability of check stated in this sentence ? | 02:02 |
blackburn | using this preproc | 02:03 |
serialhex | what do you mean? the option to not do the check and maybe have everything mess up on you? | 02:03 |
blackburn | no option | 02:03 |
blackburn | there is no check | 02:03 |
blackburn | so it is something like caution to user | 02:03 |
serialhex | oh, there _is_ no check? | 02:04 |
blackburn | there are? :D | 02:04 |
blackburn | hmm | 02:04 |
serialhex | oh, ok... umm... hd on then (i thought there was a check :P ) | 02:04 |
blackburn | yeah | 02:04 |
blackburn | that is I'm talking about | 02:04 |
blackburn | As no connectivity check is provided ..? | 02:04 |
blackburn | may be this way? | 02:04 |
serialhex | https://gist.github.com/50837ac8b233fa396be1 | 02:05 |
serialhex | or as... but that should be sufficient | 02:05 |
blackburn | ah! | 02:06 |
blackburn | yest | 02:06 |
blackburn | yes | 02:06 |
serialhex | ok cool!! :D | 02:06 |
blackburn | thanks a lot | 02:06 |
blackburn | we've finished | 02:07 |
serialhex | YAY!!! | 02:07 |
serialhex | like i said, let me know through e-mail if there are any more you want to do | 02:07 |
blackburn | okay | 02:07 |
CIA-87 | shogun: Sergey Lisitsyn master * re271fc6 / (6 files): Fixed mistakes in doc of dim reduction preprocessors - http://bit.ly/qZrtHQ | 02:07 |
serialhex | (it'll be better for you & me being in different time-zones & all) | 02:07 |
blackburn | now I have no shame sending it to google haha | 02:08 |
serialhex | :) | 02:08 |
serialhex | alright, well you probably need some sleep and i need to do some work, so i'll catch ya later! | 02:10 |
blackburn | see you | 02:10 |
blackburn | thanks again | 02:10 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 02:10 | |
-!- in3xes [~in3xes@27.107.144.35] has joined #shogun | 08:34 | |
-!- in3xes [~in3xes@27.107.144.35] has quit [Ping timeout: 264 seconds] | 09:05 | |
-!- in3xes [~in3xes@27.107.144.35] has joined #shogun | 09:06 | |
-!- in3xes [~in3xes@27.107.144.35] has quit [Remote host closed the connection] | 10:17 | |
CIA-87 | shogun: Soeren Sonnenburg master * rd2662fe / src/interfaces/python_modular/swig_typemaps.i : fix SGNDArray python typemap - http://bit.ly/rpHQ7p | 10:19 |
CIA-87 | shogun: Soeren Sonnenburg master * r9761262 / src/shogun/classifier/vw/VwEnvironment.cpp : rename lenght -> len to fix warning about naming collision - http://bit.ly/qIwMmo | 10:19 |
CIA-87 | shogun: Soeren Sonnenburg master * reb4c839 / (5 files in 2 dirs): start to transition structure to SG* data types - http://bit.ly/njVmaN | 10:19 |
-!- heiko [~heiko@134.91.53.112] has joined #shogun | 10:25 | |
@sonney2k | heiko, good morning | 10:27 |
heiko | sonney2k, hi | 10:27 |
heiko | sonney2k, I will go for the bug now immediately | 10:28 |
@sonney2k | heiko, apologies for dragging you into parameter / serialization business. it seems it is becoming hefty now | 10:28 |
@sonney2k | heiko, please read my email... | 10:28 |
heiko | sonney2k, ok, just read the mail | 10:30 |
heiko | kind of weird bug, that pops up when a method is added | 10:30 |
@sonney2k | heiko, lets see it was badly wrong before and we were just lucky that it worked | 10:31 |
@sonney2k | *extremely* lucky | 10:31 |
@sonney2k | it took me ages to figure out that adding virtual is the problem and why | 10:32 |
@sonney2k | (wouldn't have been possible w/o git bisect) | 10:32 |
heiko | yes, this is not the std bug, double free thing | 10:32 |
@sonney2k | heiko, now we at least know what the problem is | 10:33 |
@sonney2k | heiko, any ideas how to solve it? | 10:33 |
heiko | sonney2k, i am currently reading the mail again to know what exactly happens | 10:34 |
@sonney2k | I currently only see that one would have to introduce a special type for SG{vector,matrix,...} etc | 10:34 |
heiko | does this happen at every serialization of SGVector? | 10:34 |
@sonney2k | otherwise you cannot distinguish double* x, int len from SGVector<double>(x) | 10:34 |
@sonney2k | and so wen de-serializing -> kaboom | 10:34 |
@sonney2k | heiko, of course | 10:35 |
@sonney2k | heiko, de-serialization | 10:35 |
heiko | yes, i agree, a new type would fix the problem | 10:35 |
@sonney2k | heiko, I removed the virtual for now to be able to enable the test suite soon | 10:35 |
@sonney2k | (SG transition in structure is needed) | 10:38 |
@sonney2k | then all examples run through at least | 10:38 |
heiko | SG transition? | 10:39 |
heiko | ah ok, so we do this bug later and the version transition first? | 10:39 |
@sonney2k | heiko, I do this transition | 10:41 |
@sonney2k | you do the bug please :) | 10:41 |
heiko | should I send you the stuff I already did? | 10:41 |
heiko | I did a routine for loading all parameters of a file into a DynArray (by mapping all current parameters backwards until file version is reached) | 10:42 |
@sonney2k | I feel really bad about putting you into all this trouble but we really really need this serialization to be able to check for regression tests and release version 1.0 ... | 10:42 |
@sonney2k | heiko, that is the unrelated migration stuff right? | 10:42 |
@sonney2k | heiko, maybe we have a misunderstanding | 10:43 |
@sonney2k | I will do the SG transition of structure | 10:43 |
heiko | ? | 10:43 |
@sonney2k | you please do the bug / migration | 10:43 |
heiko | ah ok | 10:43 |
heiko | what is transition of structure then? | 10:44 |
@sonney2k | shogun/structure/* | 10:44 |
@sonney2k | double* x len int -> SGVector | 10:44 |
heiko | alright | 10:48 |
heiko | sonney2k, sorry, but this double* x len int -> SGVector, are you going to do it or me? | 10:49 |
@sonney2k | *me* | 10:49 |
heiko | ok | 10:50 |
heiko | but what is the bug then? :) | 10:50 |
@sonney2k | as I said I do the structure conversion | 10:50 |
@sonney2k | ?!? | 10:50 |
heiko | (10:43:38) sonney2k: you please do the bug / migration | 10:50 |
@sonney2k | the bug I mentioned in the email | 10:50 |
@sonney2k | s | 10:50 |
@sonney2k | and the serialization version migration | 10:50 |
heiko | ok | 10:50 |
heiko | alright, finally :) | 10:51 |
@sonney2k | heiko, so just to reassure me what exactly are you doing now (or planning to do?) | 10:52 |
heiko | Have a loook on the bug, with the SGFREE of a virutal method | 10:53 |
heiko | then continue on the parameter version migration | 10:53 |
* sonney2k is relieved :) | 10:53 | |
@sonney2k | for the 'SGFREE' bug: you will need to introduce new types I am afraid | 10:54 |
heiko | yes, I think so | 10:55 |
@sonney2k | heiko, and to trigger it again just git revert 49ff85135f6e96fe2527846e316f85f02f6980ca | 10:55 |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 11:21 | |
heiko | sonney2k, i have a question | 11:42 |
heiko | this svmlight serial example works via pickle | 11:42 |
heiko | where is the connection to shogun's save_serializable methods? | 11:42 |
blackburn | heiko: it uses save_serializable to temporary file then pickle | 11:46 |
blackburn | not really a pickle | 11:46 |
heiko | where is this done in the code? | 11:47 |
blackburn | at the bottom of interfaces/modular/SGBase.i | 11:48 |
heiko | thanks | 11:48 |
-!- heiko [~heiko@134.91.53.112] has quit [Quit: Leaving.] | 12:02 | |
-!- heiko1 [~heiko@134.91.53.112] has joined #shogun | 12:02 | |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 12:33 | |
@sonney2k | heiko1, if you want to test it you could just save_serializable / load_serializable of a sgvector | 12:48 |
@sonney2k | heiko1, you don't need the 'big' python example | 12:48 |
heiko1 | sonney2k, yes i am actually doing this :) | 13:27 |
@sonney2k | ok ok... | 13:28 |
heiko1 | sonney2k, the save_serializeable also does not work | 13:30 |
heiko1 | I think because the adress of the virtual method is written casted as double | 13:31 |
@sonney2k | heiko1, yes | 13:31 |
heiko1 | file contains crap | 13:31 |
@sonney2k | makes sense | 13:31 |
heiko1 | sonney2k, will we have any vectors/matrices with in the "old" style after your transition? | 13:32 |
@sonney2k | heiko1, we cannot complete the transition within the next few weeks | 13:33 |
heiko1 | sonney2k, ok then we need the old stuff to work AND the new stuff | 13:33 |
@sonney2k | so yes there still will be oldstyle crap for quite a while | 13:34 |
heiko1 | so indeed, we need new types | 13:34 |
@sonney2k | they are needed anyways | 13:34 |
@sonney2k | or how would you distinguish an old file from a new one? | 13:34 |
heiko1 | this could perhaps be done by this parameter version stuff, but ok | 13:35 |
heiko1 | n ew types are easier | 13:35 |
heiko1 | sonney2k, so where would you add these types? | 13:37 |
heiko1 | in the containerType enum? | 13:37 |
@sonney2k | makes sense to me | 13:41 |
@sonney2k | certainly not primitive type | 13:41 |
heiko1 | ok, then I will simply add two new types for container type | 13:41 |
heiko1 | SGVECTOR and SGMATRIX or so | 13:41 |
heiko1 | and add them to all switch cases | 13:42 |
heiko1 | sonney2k, then it should work, basically the casts have to be changed in the new code | 13:42 |
@sonney2k | OK | 13:42 |
heiko1 | sonney2k, ok hope it will work ... | 13:43 |
@sonney2k | so we just have a few extra case's with SGVECTOR/MATRIX/SGSTRINGLIST etc | 13:43 |
@sonney2k | heiko1, try for SGVector only first | 13:43 |
heiko1 | sonney2k, yes i was thinking that | 13:43 |
@sonney2k | heiko1, at some point we can hopefully deprecate the old double* etc stuff | 13:44 |
@sonney2k | I think then we even don't need EStructType | 13:45 |
@sonney2k | and things can be consolidated quite a bit | 13:45 |
@sonney2k | anyways for now adding cases is the only viable option | 13:45 |
@sonney2k | heiko1, so it is good that you warn on every old file now :) | 13:51 |
@sonney2k | this will force users to upgrade soon | 13:51 |
heiko1 | hehe :) | 13:52 |
-!- in3xes [~in3xes@180.149.49.227] has joined #shogun | 15:33 | |
-!- f-x [~user@117.192.204.79] has joined #shogun | 15:40 | |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 16:03 | |
-!- f-x [~user@117.192.204.79] has quit [Remote host closed the connection] | 16:05 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 258 seconds] | 16:07 | |
-!- f-x [~user@117.192.204.79] has joined #shogun | 16:33 | |
--- Log opened Wed Aug 17 19:54:40 2011 | ||
-!- shogun-toolbox [~shogun@7nn.de] has joined #shogun | 19:54 | |
-!- Irssi: #shogun: Total of 6 nicks [0 ops, 0 halfops, 0 voices, 6 normal] | 19:54 | |
-!- Irssi: Join to #shogun was synced in 6 secs | 19:54 | |
-!- sonney2k [~shogun@7nn.de] has joined #shogun | 19:54 | |
-!- mode/#shogun [+o sonney2k] by ChanServ | 19:54 | |
-!- You're now known as Guest86906 | 19:55 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 19:57 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 20:00 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:00 | |
-!- You're now known as shogun-toolbox-l | 20:02 | |
@sonney2k | f-x, why did you remove streamingfilefrom* classes from the .i file? | 20:08 |
f-x | sonney2k: just checking how much difference they made in the size | 20:09 |
@sonney2k | f-x, almost nothing I would say | 20:09 |
f-x | yeah, very little difference | 20:09 |
f-x | sonney2k: i'll re-add them | 20:09 |
-!- You're now known as shogun-irclog | 20:21 | |
@sonney2k | f-x, done? | 20:29 |
f-x | sonney2k: not yet, was doing something else.. i'll do it within 5 minutes | 20:30 |
f-x | sonney2k: now done | 20:38 |
in3xes | that is 8 mins :P | 20:38 |
CIA-87 | shogun: Shashwat Lal Das master * r79ae518 / (5 files): Added VW to swig definitions. - http://bit.ly/rpUNNA | 20:40 |
CIA-87 | shogun: Shashwat Lal Das master * red771b8 / (2 files): A couple of documentation additions. - http://bit.ly/rfTF6g | 20:40 |
CIA-87 | shogun: Shashwat Lal Das master * rf851c74 / src/interfaces/modular/IO.i : Readded StreamingFileFromFeatures to modular. - http://bit.ly/oxampZ | 20:40 |
CIA-87 | shogun: Soeren Sonnenburg master * rb477e8d / (7 files in 2 dirs): | 20:40 |
CIA-87 | shogun: Merge pull request #301 from frx/streaming_vw | 20:40 |
CIA-87 | shogun: VW in modular interfaces - http://bit.ly/qDmfPz | 20:40 |
-!- in3xes_ [~in3xes@180.149.49.227] has joined #shogun | 20:59 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Ping timeout: 240 seconds] | 21:02 | |
-!- in3xes_ is now known as in3xes | 21:02 | |
-!- f-x [~user@117.192.204.79] has quit [Ping timeout: 260 seconds] | 21:38 | |
-!- in3xes [~in3xes@180.149.49.227] has quit [Quit: Leaving] | 21:44 | |
CIA-87 | shogun: Soeren Sonnenburg master * ra73d0cb / (5 files in 2 dirs): transition further parts to SG* datatypes in structure/ - http://bit.ly/oKLBwE | 22:04 |
-!- f-x [~user@117.192.194.115] has joined #shogun | 22:05 | |
blackburn | sonney2k: what's up with serialization bug? | 22:30 |
@sonney2k | blackburn, heiko didn't commit yet | 22:31 |
@sonney2k | but he is close to having resolved it | 22:31 |
blackburn | nice | 22:31 |
@sonney2k | blackburn, I converted most of the structure/* stuff to use sgvector but it crashes... | 22:31 |
blackburn | will remove some more warnings | 22:31 |
@sonney2k | not my code | 22:31 |
blackburn | uh.. | 22:31 |
blackburn | what crashes? | 22:32 |
@sonney2k | so I don't really understand why it is crashing | 22:32 |
@sonney2k | (and the octave_static interface using that code still works!) | 22:32 |
@sonney2k | my machine is still pretty unstable... ohh well | 22:33 |
@sonney2k | I better sleep now | 22:33 |
blackburn | see you.. | 22:33 |
CIA-87 | shogun: Sergey Lisitsyn master * r36676f3 / src/shogun/ui/SGInterface.h : SGInterface doxygen warning removal - http://bit.ly/p9Jxs3 | 22:52 |
CIA-87 | shogun: Soeren Sonnenburg master * r3d3eae4 / src/NEWS : mention that we support matlab 2011a now too - http://bit.ly/pVhj8E | 23:05 |
CIA-87 | shogun: Soeren Sonnenburg master * r636c1fa / src/shogun/mathematics/Math.h : fix warning in logsum_array - http://bit.ly/plO1v3 | 23:05 |
blackburn | strange sleep | 23:07 |
blackburn | haha | 23:07 |
@sonney2k | you are talking to a zombie | 23:12 |
CIA-87 | shogun: Soeren Sonnenburg master * r96da4ad / (2 files in 2 dirs): fix compile errors detected by nightly_all - http://bit.ly/qPIa2K | 23:31 |
CIA-87 | shogun: Soeren Sonnenburg master * r2a787b9 / src/NEWS : Merge branch 'master' of ssh://shogun-toolbox.org/shogun - http://bit.ly/pe3JXW | 23:33 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 23:36 | |
--- Log closed Thu Aug 18 00:00:33 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!