| --- Log opened Sat Jun 14 00:00:32 2014 | ||
| -!- zxtx_ [~zv@host109-147-162-17.range109-147.btcentralplus.com] has joined #shogun | 00:29 | |
| -!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 00:40 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 00:40 | |
| -!- zxtx_ [~zv@host109-147-162-17.range109-147.btcentralplus.com] has quit [Ping timeout: 240 seconds] | 01:34 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has joined #shogun | 01:36 | |
| thoralf | Hey. | 01:37 |
|---|---|---|
| -!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Ping timeout: 252 seconds] | 02:25 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has quit [Quit: Konversation terminated!] | 02:29 | |
| shogun-buildbot | build #27 of nightly_fedora is complete: Failure [failed git] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_fedora/builds/27 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Parijat Mazumdar <mazumdarparijat@gmail.com> | 03:00 |
| -!- soumyaC [uid15286@gateway/web/irccloud.com/x-qrkukgbnezblckhi] has quit [Quit: Connection closed for inactivity] | 05:56 | |
| shogun-buildbot | build #826 of nightly_default is complete: Failure [failed notebooks] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/826 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, lambday <heavensdevil6909@gmail.com>, Fernando Iglesias <fernando.iglesiasg@gmail.com>, Parijat Mazumdar <mazumdarparijat@gmail.com> | 07:40 |
| -!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 07:52 | |
| -!- mode/#shogun [+o iglesiasg] by ChanServ | 07:52 | |
| -!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving] | 08:13 | |
| -!- kislay [~Abhijeet@103.240.205.45] has joined #shogun | 08:25 | |
| -!- abinash_panda [~abinash@117.200.235.73] has joined #shogun | 13:47 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has joined #shogun | 15:08 | |
| thoralf | Heyhey. | 15:34 |
| thoralf | abinash_panda: Hey. | 15:36 |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has quit [Ping timeout: 252 seconds] | 16:15 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has joined #shogun | 16:21 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has quit [Ping timeout: 240 seconds] | 17:13 | |
| -!- abinash_panda [~abinash@117.200.235.73] has quit [Ping timeout: 240 seconds] | 18:53 | |
| -!- abinash_panda [~abinash@223.179.142.166] has joined #shogun | 19:11 | |
| -!- zxtx_ [~zv@31.55.63.253] has joined #shogun | 19:25 | |
| -!- soumyaC [uid15286@gateway/web/irccloud.com/x-qmsfqbyijxobnixe] has joined #shogun | 20:04 | |
| -!- zxtx_ [~zv@31.55.63.253] has quit [Ping timeout: 245 seconds] | 20:11 | |
| -!- khalednasr [~k.nasr92@41.69.194.156] has joined #shogun | 20:52 | |
| -!- abinash_1anda [~abinash@110.227.152.138] has joined #shogun | 20:54 | |
| -!- abinash_panda [~abinash@223.179.142.166] has quit [Ping timeout: 244 seconds] | 20:57 | |
| -!- kislay [~Abhijeet@103.240.205.45] has quit [Ping timeout: 240 seconds] | 20:59 | |
| -!- abinash_1anda is now known as abinash_panda | 21:00 | |
| -!- thoralf [~thoralf@91-65-141-142-dynip.superkabel.de] has joined #shogun | 21:15 | |
| abinash_panda | hey thoralf ! | 21:15 |
| thoralf | Hey abinash_panda | 21:15 |
| abinash_panda | I have some doubts regarding the hashing of sparse vector. | 21:16 |
| abinash_panda | I went through the paper K. Weinberger, et.al. Feature Hashing for Large Scale Multitask Learning | 21:16 |
| abinash_panda | and have some confusion regarding the hashing | 21:16 |
| abinash_panda | I had sent you a mail regarding it ;) | 21:16 |
| thoralf | I know - but I don't understand what's the difference between both formulas. | 21:17 |
| thoralf | Ah, I see. | 21:18 |
| thoralf | No, the formula is correct. | 21:18 |
| thoralf | You want wo sum over all indices, which are hashed to the same value. | 21:18 |
| abinash_panda | thoralf: Okay. Thats what I am saying that we should sum over all the x[j] j:h(j) == i | 21:20 |
| abinash_panda | instead of x[i] j:h(j) == i | 21:20 |
| thoralf | \xi(j) x_j | 21:21 |
| abinash_panda | or \xi(i) x_j ?? | 21:22 |
| thoralf | I see your point. It's a typo in the paper. | 21:22 |
| thoralf | We have two hashing functions over (j). | 21:22 |
| thoralf | One is hashing the feature indices, the other is "toggling" the prefix of the summed value. | 21:23 |
| abinash_panda | Okay. I got it. | 21:24 |
| thoralf | alex.smola.org/papers/2009/Weinbergeretal09.pdf | 21:24 |
| thoralf | They fixed it in this paper. | 21:24 |
| abinash_panda | So, with this I had modified my hash_vector() method and the results are | 21:24 |
| abinash_panda | [0.0 sec] created random sparse vector: 262144 features, 268435456 dimensions, [min;max;len]=[440;268435454;261995] | 21:25 |
| thoralf | But I found out that your version is slower. | 21:25 |
| abinash_panda | [10.7 sec] tklein23 implementation -- 24.0 | 21:25 |
| abinash_panda | [3.2 sec] abinash implementation -- 79.8 | 21:25 |
| thoralf | You need to create "sparse" hashed vectors. | 21:25 |
| thoralf | uint32_t hash_dim = 1 << 18 | 21:25 |
| thoralf | int32_t num_features = 1 << 12 | 21:25 |
| thoralf | Both parameters were switched, thus creating *dense* vectors. | 21:26 |
| thoralf | Your implementation always creates a *dense* vector as intermediate vector. | 21:27 |
| thoralf | It has to be allocated, zero()ed, assigned and then iterated over. | 21:27 |
| thoralf | And if the hash space is big enough, this can be expensive. | 21:27 |
| -!- lisitsyn [~qdrgsm@85.114.187.192] has joined #shogun | 21:30 | |
| lisitsyn | khalednasr: hey | 21:30 |
| abinash_panda | Yeah, I have seen it. | 21:30 |
| thoralf | Master! | 21:30 |
| khalednasr | lisitsyn, hey | 21:30 |
| lisitsyn | khalednasr: gonna glance over your pr just now | 21:30 |
| lisitsyn | thoralf: hah | 21:31 |
| khalednasr | lisitsyn, thanks! take your time | 21:31 |
| abinash_panda | thoralf: One minor modification required in your code: | 21:31 |
| abinash_panda | [0.0 sec] created random sparse vector: 262144 features, 268435456 dimensions, [min;max;len]=[440;268435454;261995] | 21:31 |
| abinash_panda | [10.7 sec] tklein23 implementation -- 24.0 | 21:31 |
| abinash_panda | dst.features[j].entry += (mm3 % 2 == 1 ? -1.0 : 1.0) * src.features[j].entry; | 21:31 |
| abinash_panda | thoralf: dst.features[j].entry += (mm3 % 2 == 1 ? -1.0 : 1.0) * src.features[j].entry; | 21:31 |
| lisitsyn | yikes! 268435456 dimensions ;) | 21:32 |
| thoralf | lisitsyn: :D | 21:33 |
| thoralf | abinash_panda: I don't see the improvement. | 21:33 |
| thoralf | += | 21:33 |
| thoralf | Damn! | 21:33 |
| thoralf | Yes | 21:33 |
| lisitsyn | khalednasr: as usual no comments :D | 21:33 |
| -!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 21:33 | |
| shogun-notifier- | shogun: khalednasr :develop * d254dbb / / (5 files): https://github.com/shogun-toolbox/shogun/commit/d254dbb51a1f8440bff334e769766a3fd6648753 | 21:33 |
| shogun-notifier- | shogun: Implemented Restricted Boltzmann Machines | 21:33 |
| shogun-notifier- | shogun: khalednasr :develop * 12ce633 / NEWS: https://github.com/shogun-toolbox/shogun/commit/12ce633d01841d3da9a5742f5c6808589870daf5 | 21:33 |
| shogun-notifier- | shogun: updated NEWS file to include changes to the NeuralNets module | 21:33 |
| shogun-notifier- | shogun: Sergey Lisitsyn :develop * 2490b57 / / (6 files): https://github.com/shogun-toolbox/shogun/commit/2490b576b8079aee59ef3074357d3b30ba9daebc | 21:33 |
| shogun-notifier- | shogun: Merge pull request #2304 from khalednasr/rbms | 21:33 |
| shogun-notifier- | shogun: | 21:33 |
| shogun-notifier- | shogun: Implemented Restricted Boltzmann Machines | 21:33 |
| thoralf | abinash_panda: No! | 21:34 |
| thoralf | :D | 21:34 |
| thoralf | abinash_panda: I'm only collecting the values. | 21:34 |
| lisitsyn | it is boring you write that good code :P | 21:34 |
| thoralf | abinash_panda: The summing happens magically when sort_features() | 21:34 |
| khalednasr | lisitsyn, haha :D | 21:34 |
| thoralf | abinash_panda: If a sparse vector has duplicate entries, they will be summed after sorting. | 21:35 |
| abinash_panda | thoralf: Okay. I was not knowing it. Cool feature ;) | 21:36 |
| abinash_panda | thoralf: Okay. So, I am going with your implementation of hash_vector() :) | 21:36 |
| thoralf | lol | 21:36 |
| thoralf | abinash_panda: Well, we have to take care that sparse vectors indices are sorted and not duplicate. | 21:37 |
| thoralf | abinash_panda: Otherwise strange[tm] things will happen. | 21:37 |
| abinash_panda | Yeah...BTW I liked the *strange[tm]* comment :D | 21:39 |
| thoralf | :D | 21:39 |
| abinash_panda | thoralf: One more thing, if we are going to deal only with SGSparseVector, so should I put a check on features being CSparseFeatures ? | 21:40 |
| thoralf | abinash_panda: Good question. | 21:41 |
| thoralf | abinash_panda: Yes, please check it. | 21:42 |
| abinash_panda | Yeah...then we should also go for CHashedSparseMultilabelModel to avoid confusion ;) | 21:42 |
| * thoralf hates these long names. | 21:43 | |
| thoralf | :D | 21:43 |
| shogun-buildbot | build #2373 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2373 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com>, khalednasr <k.nasr92@gmail.com> | 21:43 |
| abinash_panda | thoralf: Fine. Then I am going with CHashedMultilabelModel :D | 21:43 |
| thoralf | Other thing is that we add alternate implemation for hash_vector() and put a SGNOTIMPLEMENTED inside. | 21:44 |
| thoralf | abinash_panda: Haha, cool. | 21:44 |
| thoralf | You know how to make people happy ;) | 21:44 |
| abinash_panda | thoralf: Yeah ;) | 21:45 |
| abinash_panda | thoralf: another implementation for hash_vector()?? | 21:45 |
| abinash_panda | thoralf: is this for dense vectors?? | 21:45 |
| thoralf | abinash_panda: Not yes, just a bold idea. | 21:45 |
| thoralf | sth like hash_vector(SGVector) and hash_vector(SGSparseVector). | 21:46 |
| thoralf | But don't care about this yet. I don't see a sense in using *dense* vectors with this approach. | 21:46 |
| thoralf | This will eat up your memory, your family and your whole life. | 21:47 |
| abinash_panda | thoralf: lol :D | 21:47 |
| thoralf | abinash_panda: Have you tried the benchmark with hash_dim = 1 << 18 and num_features = 1 << 12 ? | 21:48 |
| abinash_panda | Yeah, I had tried. Found that your implementation is faster ;) | 21:50 |
| thoralf | Okay. | 21:50 |
| thoralf | Difference will grow with increaing hash_dim. | 21:50 |
| thoralf | 1 << 20 is devastating | 21:50 |
| abinash_panda | Oh. No, I have not tried yet. | 21:51 |
| abinash_panda | thoralf: Another doubt regarding get_joint_feature_vector() method? | 21:51 |
| thoralf | abinash_panda: Do you have doubts or are you asking me? | 21:52 |
| abinash_panda | As you have mentioned in the comment that it should be a sparse vector right? | 21:52 |
| thoralf | abinash_panda: Yes. | 21:53 |
| -!- pickle27 [~pickle27@192-0-136-118.cpe.teksavvy.com] has joined #shogun | 21:53 | |
| thoralf | abinash_panda: My local code already uses sparse vectors for this, but it's hard to isolate... | 21:54 |
| abinash_panda | thoralf: But the CStructuredModel class requires that the output of this should be SGVector instead. | 21:54 |
| thoralf | abinash_panda: I know. Let me check something. | 21:55 |
| abinash_panda | thoralf: Furthermore, solvers like PrimalMosekSOSVM and StochasticSOSVM also require ret->psi_pred and ret->psi_truth | 21:55 |
| thoralf | abinash_panda: In my refactoring I completely removed psi_pred and psi_truth from result set. | 21:56 |
| abinash_panda | Okay. | 21:57 |
| thoralf | Instead I put get_joint_feature_vector calls directly in the machines. | 21:57 |
| abinash_panda | thoralf: Yeah...we can do that | 21:58 |
| thoralf | abinash_panda: We should. | 21:58 |
| thoralf | Otherwise we'll get into trouble with our approach. | 21:58 |
| thoralf | Can you make it a separate PR and send it before the current one? | 21:59 |
| abinash_panda | Yeah..sure. | 21:59 |
| thoralf | iirc this is only small refactoring - but I don't remember if I had to fix unit tests as well. | 21:59 |
| abinash_panda | Once you were talking of some good git tutorial. Could you please send me the link? | 22:00 |
| abinash_panda | I had messed up with my local repo | 22:00 |
| thoralf | abinash_panda: Mail sent. | 22:01 |
| thoralf | abinash_panda: It's a paid one, so please don't spread it too far. | 22:01 |
| abinash_panda | thoralf: Okay. I would take care of that. | 22:02 |
| abinash_panda | Thanks :) | 22:02 |
| abinash_panda | thoralf: So, tomorrow I would sending the PR refactoring the solvers :) | 22:04 |
| thoralf | The machines, right? | 22:04 |
| thoralf | Not the solvers | 22:04 |
| thoralf | I grepped for psi_truth/psi_pred and found a few machines. | 22:05 |
| abinash_panda | Yeah..the machines. | 22:05 |
| thoralf | So first refactoring is simply removing psi_pred/truth from ResultSet and calling psi() inside the machines. | 22:05 |
| abinash_panda | Yeah...it would be a simple task. I would do it. | 22:07 |
| abinash_panda | and then we can discuss on the get_joint_feature_vector() | 22:07 |
| thoralf | Yes. | 22:07 |
| abinash_panda | we can overload get_joint_feature_vector() method for returning SGSparseVector..right? | 22:08 |
| thoralf | abinash_panda: Not sure if we can have same signature but different return values. | 22:09 |
| thoralf | lisitsyn: Master? | 22:09 |
| lisitsyn | thoralf: ahh stop it :D | 22:09 |
| thoralf | :D | 22:09 |
| thoralf | lisitsyn: X function(Y) can't have a Z function() implementation, right? | 22:09 |
| thoralf | X function(Y) | 22:09 |
| lisitsyn | thoralf: it has some intrinsic irony and I feel I don't deserve being msc :D | 22:10 |
| thoralf | Z function(Y) | 22:10 |
| lisitsyn | ehhm | 22:10 |
| lisitsyn | ah covariant types? | 22:10 |
| lisitsyn | or? | 22:10 |
| thoralf | Don't know what covariant types are. :D | 22:10 |
| lisitsyn | ok in java you can | 22:10 |
| lisitsyn | Z extends X | 22:10 |
| lisitsyn | then | 22:10 |
| thoralf | No | 22:10 |
| lisitsyn | totally different types? | 22:11 |
| thoralf | Z extends A and X extends A | 22:11 |
| thoralf | A = SGObject ;) | 22:11 |
| thoralf | so totally different | 22:11 |
| * thoralf cannot imaging how dispatching to the right function should work. ;) | 22:11 | |
| lisitsyn | how can it dispatch what function to call? ;) | 22:12 |
| lisitsyn | well | 22:12 |
| thoralf | imagine | 22:12 |
| thoralf | Haha, thanks. | 22:12 |
| thoralf | abinash_panda: Let's say we can't - but let's discuss that tmrw | 22:12 |
| lisitsyn | in C++ you can | 22:12 |
| thoralf | lisitsyn: Really? | 22:12 |
| abinash_panda | thoralf: Fine :) | 22:12 |
| lisitsyn | yeah with default argument | 22:13 |
| lisitsyn | it looks like | 22:13 |
| lisitsyn | X function(Y, somedefault thing) | 22:13 |
| lisitsyn | Z function(Y, some default thing) | 22:13 |
| lisitsyn | the trick is you need to FAIL one default thing | 22:14 |
| lisitsyn | so it has no other choice | 22:14 |
| thoralf | And how do I fail? | 22:14 |
| lisitsyn | hmm accessing some member of class | 22:15 |
| lisitsyn | I mean if you ask for | 22:15 |
| lisitsyn | A::T | 22:15 |
| lisitsyn | and there is no T | 22:15 |
| lisitsyn | in A | 22:15 |
| lisitsyn | it fails | 22:15 |
| lisitsyn | http://en.wikibooks.org/wiki/More_C%2B%2B_Idioms/SFINAE | 22:15 |
| lisitsyn | ok this describes better | 22:15 |
| thoralf | Ah! function templates | 22:16 |
| thoralf | Of course. | 22:16 |
| thoralf | Not overloading, but templates. | 22:16 |
| lisitsyn | thoralf: yeah sure | 22:16 |
| thoralf | lisitsyn: In shogun we're using everything. ;) | 22:16 |
| lisitsyn | thoralf: but in normal languages there is no such crazy thing as overloaded return type functions | 22:17 |
| thoralf | lisitsyn: Are you calling C++ crazy? ;) | 22:17 |
| lisitsyn | thoralf: absolutely | 22:17 |
| thoralf | Agree :) | 22:17 |
| thoralf | C++ is kind of a job guarantee. :) | 22:18 |
| lisitsyn | thoralf: though java sucks | 22:18 |
| lisitsyn | :D | 22:18 |
| thoralf | (Java is no job guarantee itself; you need to add JavaEE or Hadoop to make it unpredictable ;)) | 22:18 |
| lisitsyn | thoralf: on the other hand java is too verbose and is no better | 22:18 |
| lisitsyn | hmm here I had a lot opportunities with java | 22:19 |
| lisitsyn | but no opportunities with C++ at all | 22:19 |
| thoralf | opportunities? | 22:19 |
| thoralf | Talking seriously or ironic? | 22:19 |
| lisitsyn | well | 22:19 |
| lisitsyn | no, I am quite serious | 22:19 |
| thoralf | opportunities always sounds like "issues" ;) | 22:19 |
| lisitsyn | when I wanted to change my job | 22:19 |
| lisitsyn | I had no C++ thing around | 22:20 |
| @wiking | lisitsyn: so yeah i'm moving to .sg | 22:20 |
| thoralf | Yeah, but Java itself is easy to learn. That's why all the companies are using Java, because developers are easier to find. | 22:20 |
| lisitsyn | wiking: sg! | 22:20 |
| thoralf | C++ is hard to learn, so less people learn it :) | 22:20 |
| lisitsyn | thoralf: yeah monkey can code java | 22:21 |
| lisitsyn | :D | 22:21 |
| thoralf | wiking: Cool. | 22:21 |
| thoralf | wiking: Don't do anything wrong - they may put you in chains and whip you. ;) | 22:21 |
| lisitsyn | wiking: how do you feel about it? ;) | 22:21 |
| @wiking | thoralf: hhahhahah | 22:22 |
| @wiking | lisitsyn: well a bit ambivalent | 22:23 |
| @wiking | but it's ok i always wanted to live in asia | 22:23 |
| @wiking | so we'll see | 22:23 |
| lisitsyn | shogunees move around the world | 22:23 |
| lisitsyn | :D | 22:23 |
| lisitsyn | yaa! | 22:23 |
| thoralf | wiking: Wish you good luck. | 22:23 |
| @wiking | but yeah after the move | 22:23 |
| lisitsyn | wiking: when do you move? | 22:23 |
| @wiking | i'll be back in working on shogun | 22:23 |
| @wiking | lisitsyn: hehehehe i thought it's going to be 1st of august | 22:23 |
| thoralf | wiking:) | 22:24 |
| lisitsyn | wiking: cool we are moving at the same time hah | 22:24 |
| @wiking | but now my wife checked an email and suddenly it seems we have to go on the 9th of july | 22:24 |
| @wiking | :))) | 22:24 |
| lisitsyn | ah | 22:24 |
| @wiking | so suddenly it's like getting tickets | 22:24 |
| pickle27 | wiking: where are you moving to? | 22:24 |
| @wiking | pickle27: singapore | 22:24 |
| pickle27 | ahh cool | 22:24 |
| pickle27 | I know some guys who have worked there | 22:24 |
| pickle27 | they really like it | 22:24 |
| @wiking | ok i need job | 22:24 |
| @wiking | :) | 22:24 |
| @wiking | so anybody knows anything in singapore | 22:25 |
| @wiking | plz let me know | 22:25 |
| @wiking | as currently i dont get to work in singapore (visa problems) | 22:25 |
| @wiking | so i need an employer who would apply for me for a working visa :P | 22:25 |
| pickle27 | I still think you guys should all come work for shopify :p | 22:25 |
| -!- travis-ci [~travis-ci@ec2-50-16-123-177.compute-1.amazonaws.com] has joined #shogun | 22:25 | |
| travis-ci | [travis-ci] it's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/27582468 | 22:25 |
| -!- travis-ci [~travis-ci@ec2-50-16-123-177.compute-1.amazonaws.com] has left #shogun [] | 22:25 | |
| lisitsyn | pickle27: give me visa I am all yours haha | 22:25 |
| @wiking | so yeah | 22:25 |
| @wiking | i'm for sale at the moment | 22:25 |
| @wiking | as i had to quit my current job | 22:25 |
| @wiking | because of the move | 22:26 |
| @wiking | pickle27: shopify? work from remote? :) | 22:26 |
| lisitsyn | pickle27: you probably don't even imagine what's the canadian visa PITA :D | 22:26 |
| pickle27 | we only have remote of operations at the moment | 22:26 |
| pickle27 | all devs are in canada | 22:26 |
| pickle27 | lisitsyn: I'll ask for you! | 22:27 |
| @wiking | anyhooooow thoralf if u know something at researchgate let me know :) | 22:27 |
| lisitsyn | pickle27: haha no need yet | 22:27 |
| @wiking | i'm really jobless atm | 22:27 |
| @wiking | :))) | 22:27 |
| lisitsyn | pickle27: I am moving to yandex this august | 22:27 |
| thoralf | pickle27: Canada is great. :) | 22:27 |
| lisitsyn | gonna try it first | 22:27 |
| @wiking | other than that i code for food :P | 22:27 |
| thoralf | pickle27: I had to cancel my canada vacation this week ;( | 22:27 |
| thoralf | pickle27: Shopify is looking for ML engineers? | 22:28 |
| pickle27 | oh no! thoralf where were you visiting in canada? | 22:28 |
| thoralf | pickle27: I like Canada so much and I booked a flight to Boston for last week, then going to Toronto and Montreal this week, New England next week. | 22:29 |
| thoralf | So the plan. | 22:29 |
| thoralf | But I had to cancel a day before because my gf had a bicycle accident. | 22:29 |
| lisitsyn | oh is it ok? | 22:29 |
| thoralf | She's okay-ish now. | 22:29 |
| pickle27 | that sucks hope she's alright | 22:29 |
| thoralf | Yeah, but this happened the day before the flight, so the doctors said not to fly. | 22:30 |
| thoralf | But I love canada. :D | 22:30 |
| lisitsyn | this sucks | 22:30 |
| thoralf | pickle27: Where are you living? | 22:30 |
| pickle27 | we're hiring all kinds of engineers btw we are growing too fast | 22:30 |
| pickle27 | Im in Ottawa now | 22:30 |
| pickle27 | as in we can't hire people fast enough! | 22:31 |
| thoralf | pickle27: Okay, probably we're traveling to Canada this autumn. Same tour. Indian summer. We'll let you know... you can even join, as we probably will rent a car and travel around. :D | 22:32 |
| thoralf | Boston -> Amherst -> Toronto -> ... -> Montreal -> Boston | 22:32 |
| pickle27 | yeah totally let me know! it would be fun to meet some of you guys in rl | 22:33 |
| thoralf | Will you come to Berlin for the workshop? | 22:34 |
| thoralf | Just to let you know... I can host some of you guys here. | 22:34 |
| pickle27 | I was thinking about it but I haven't planned anything so probably not | 22:34 |
| lisitsyn | thoralf: ha! this would be cool | 22:34 |
| thoralf | So no hotel needed, if this may change your decision. | 22:35 |
| lisitsyn | but I am still unavailable | 22:35 |
| thoralf | Only have a sparse sofa, but enough floor space for a lot of nerds. :D | 22:35 |
| lisitsyn | gonna rush through moscow in search of a flat exactly these days | 22:35 |
| lisitsyn | thoralf: tell me sth about winter berlin | 22:36 |
| lisitsyn | it is cool during winter? | 22:36 |
| lisitsyn | is it* :D | 22:36 |
| lisitsyn | cool not temperature wise but any other kind of coolness | 22:36 |
| thoralf | lisitsyn: I'm living here... so you shouldn't ask me. | 22:37 |
| thoralf | I don't know all this tourist stuff. | 22:37 |
| lisitsyn | thoralf: hah | 22:37 |
| thoralf | Waking up, going to work, going back. | 22:37 |
| thoralf | Spending my holiday in != Berlin ;) | 22:37 |
| lisitsyn | thoralf: are there any sales before xmas or whatever? | 22:37 |
| thoralf | lisitsyn: Of course, everywhere. | 22:38 |
| lisitsyn | this is deffs not I am interesting in but you know I have gf | 22:38 |
| lisitsyn | :D | 22:38 |
| thoralf | lisitsyn: Before christmas there are some traditional markets. | 22:38 |
| thoralf | In Berlin, Leipzig, N?rnberg... | 22:38 |
| thoralf | hot spiced wine | 22:38 |
| thoralf | :D | 22:38 |
| lisitsyn | we call it some german word | 22:39 |
| thoralf | or wine punch - however this is called. | 22:39 |
| thoralf | Gl?hwein | 22:39 |
| lisitsyn | glintwine? | 22:39 |
| lisitsyn | :D | 22:39 |
| lisitsyn | thoralf: okay I see :) | 22:40 |
| lisitsyn | thoralf: this is something funny | 22:40 |
| lisitsyn | you see you're living in kind of center of something | 22:40 |
| lisitsyn | at least more things are going on there | 22:40 |
| lisitsyn | someone is living in NYC and feel the same - wake up / work / sleep | 22:41 |
| thoralf | lisitsyn: I'm living next to Tegel Airport and Main station. | 22:41 |
| lisitsyn | what's main station? | 22:41 |
| thoralf | Hauptbahnhof. | 22:41 |
| thoralf | Train. | 22:41 |
| lisitsyn | ahh | 22:42 |
| lisitsyn | cool place | 22:42 |
| thoralf | Yeah. | 22:42 |
| lisitsyn | the only place we managed to buy things in | 22:42 |
| thoralf | Cool place for good transport experience when hosting guests. :D | 22:42 |
| lisitsyn | because I was stupid enough to not think people could ever not work during sundays | 22:42 |
| thoralf | lol | 22:42 |
| thoralf | pickle27, lisitsyn: I'll host all of you guys. :) | 22:43 |
| lisitsyn | why? :D | 22:43 |
| pickle27 | I really do want to visit Berlin! | 22:43 |
| lisitsyn | pickle27: ask me about tourist things haha ;) | 22:44 |
| thoralf | lisitsyn: You don't have to. :D | 22:44 |
| lisitsyn | thoralf: wait wait I feel cheated :D | 22:44 |
| lisitsyn | how come you live near tegel | 22:44 |
| lisitsyn | and hauptbahnhof | 22:44 |
| lisitsyn | at the same time | 22:45 |
| lisitsyn | aren't they far away? | 22:45 |
| thoralf | lisitsyn: maps.google.com -> berlin beusselstr | 22:45 |
| lisitsyn | thoralf: tegel is terribly far away! :D | 22:46 |
| thoralf | Or compute directions: Hauptbahnhof -> Beusselstr -> Tegel | 22:46 |
| thoralf | lisitsyn: It's 5-10 minutes by Bus. | 22:47 |
| thoralf | 5 by car | 22:47 |
| lisitsyn | thoralf: alright ;) | 22:47 |
| thoralf | And the bus between Hauptbahnhof and Tegel stops (almost) in front of my house. :D | 22:47 |
| lisitsyn | thoralf: this moabit thing seemed to be quiet area | 22:48 |
| lisitsyn | is it true? | 22:49 |
| thoralf | lisitsyn: It's getting popular because of last cheap flats near city center. | 22:49 |
| thoralf | lisitsyn: We have lot of students here, but (not yet) many caf?s and bars. | 22:49 |
| thoralf | But I think they're developing. | 22:50 |
| thoralf | But it's more living area for working people, but already nice enough for hipster coming here on the weekends. | 22:50 |
| lisitsyn | thoralf: ah btw I have opposite offer - I could host you in moscow from august ;) | 22:50 |
| lisitsyn | and well any of you ;) | 22:51 |
| lisitsyn | thoralf: you haven't been there right? | 22:51 |
| thoralf | lisitsyn: Honestly I'm a concerned about Russia. About war with Ukraine. | 22:51 |
| thoralf | lisitsyn: Never. | 22:51 |
| thoralf | Never been there. | 22:51 |
| lisitsyn | how concerned? | 22:52 |
| thoralf | I don't know. First of all, reputation of Russia is not quite good over here. | 22:52 |
| lisitsyn | heh well | 22:52 |
| thoralf | Regaring police and foreign affairs. | 22:52 |
| thoralf | Regarding | 22:53 |
| lisitsyn | thoralf: it is actually like not something to worry about :D | 22:54 |
| lisitsyn | I mean the best solution is to don't bother with that stuff at all | 22:54 |
| thoralf | lisitsyn: Well, I guess it doesn't affect tourists at all. | 22:55 |
| thoralf | lisitsyn: Btw., before xmas the whole city is illuminated. ;) | 22:57 |
| thoralf | (Not what you think :D) | 22:57 |
| lisitsyn | thoralf: haha | 22:58 |
| lisitsyn | thoralf: I wish this ukraine madness stopped but nobody wants to stop it | 22:59 |
| lisitsyn | thoralf: EU and the US put some sanctions but they are really insignificant | 22:59 |
| lisitsyn | so this is a big game | 22:59 |
| thoralf | lisitsyn: Everybody wants to stop it, but some people see big opportunities there... | 23:01 |
| lisitsyn | thoralf: well not everybody but all but people who control | 23:01 |
| lisitsyn | :D | 23:01 |
| lisitsyn | it's getting funny because our presidents does whatever he wants and EU has no real power to push him back | 23:02 |
| lisitsyn | president* | 23:02 |
| lisitsyn | no real power or no real will | 23:02 |
| lisitsyn | I am quite sure $-wise it is quite easy to put some costs on russia but it is not the case for some reason | 23:02 |
| thoralf | lisitsyn: Everybody here is surprised as well why we can't do anything about it. | 23:03 |
| thoralf | lisitsyn: Maybe they thing about "markets" - Russia simply is a big market and Germany lives of exports. | 23:04 |
| thoralf | And everybody else is concerned about gas? | 23:04 |
| thoralf | Only guessing. | 23:04 |
| lisitsyn | well gas is the thing, yeah | 23:04 |
| lisitsyn | it should be like 20% of gas import in the EU | 23:04 |
| lisitsyn | that comes from russia | 23:04 |
| lisitsyn | thoralf: the most important thing is that russian elite is heavily attached to the EU | 23:06 |
| lisitsyn | they live there, their children live there | 23:06 |
| lisitsyn | so this is pure fake | 23:06 |
| thoralf | They have estates in Berlin, while local elite hast estates in Moscow :D | 23:07 |
| thoralf | Just conspiracy, no proof. | 23:07 |
| lisitsyn | thoralf: and it is good | 23:08 |
| lisitsyn | :D | 23:08 |
| lisitsyn | I mean as long as they are deeply tied to europe we will not have any kind of war :D | 23:08 |
| lisitsyn | but if putin's daughters are out of netherlands we can treat it as a bad sign hah | 23:09 |
| thoralf | Good point. | 23:10 |
| thoralf | We should visualize relatives of russian politicians on a map. | 23:10 |
| thoralf | We := Western secret services | 23:11 |
| -!- lisitsyn [~qdrgsm@85.114.187.192] has quit [Read error: Connection reset by peer] | 23:11 | |
| -!- lisitsyn [~qdrgsm@85.114.187.192] has joined #shogun | 23:11 | |
| lisitsyn | thoralf: that's easy | 23:11 |
| lisitsyn | you just draw star-like topology | 23:12 |
| lisitsyn | connect moscow with every european city | 23:12 |
| lisitsyn | there is nothing outside of moscow in russia so it would look cool | 23:12 |
| thoralf | lisitsyn: Btw., have you seen memcheck on buildbot is error free now? :D | 23:17 |
| lisitsyn | thoralf: no clue! | 23:17 |
| * thoralf pushed this since march. | 23:18 | |
| lisitsyn | thoralf: oh sounds like I should congartulate you ;) | 23:18 |
| thoralf | lol | 23:18 |
| thoralf | You sound like Sheldon. | 23:19 |
| lisitsyn | thoralf: oh no I don't like that thing :D | 23:19 |
| -!- pickle27 [~pickle27@192-0-136-118.cpe.teksavvy.com] has quit [Quit: Leaving] | 23:49 | |
| --- Log closed Sun Jun 15 00:00:33 2014 | ||
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!