--- Log opened Fri Apr 20 00:00:32 2018 | ||
-shogun-buildbot:#shogun- Build nightly_bsd_license #140 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/4/builds/140 | 03:31 | |
-shogun-buildbot:#shogun- Build nightly_default #128 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/17/builds/128 | 06:31 | |
-!- sukey [~nodebot@ks312251.kimsufi.com] has joined #shogun | 10:10 | |
-!- mode/#shogun [+o sukey] by ChanServ | 10:10 | |
-!- witness [uid10044@gateway/web/irccloud.com/x-euzmyvhpfazcmxpc] has joined #shogun | 13:17 | |
@wiking | lisitsyn, would you mind if we make AnyParameter get_paramter() public? | 15:19 |
---|---|---|
@wiking | or you rather have a visit() for SGO? | 15:20 |
@wiking | nevermind | 15:20 |
@wiking | i'll move it to sgo | 15:21 |
@wiking | the whole serialization thing :) | 15:21 |
@wiking | noooo i dont wanna do interfacing !@#r4 | 15:28 |
@wiking | i must say that this is the most cumbersome design that we could have, because either one does serialization with observing the object -> need to make the deserializer and serializer a friend class of SGO or we need to have an elaborate interface for all the different serialization format :P | 15:30 |
@wiking | i dont like the observing thing really because that assumes that everything is public of the object that is required for serialization | 15:31 |
@wiking | but i dont really like writing a bunch of wrapper code just for the sake of doing it :) | 15:31 |
lisitsyn | wiking: hey | 16:45 |
@wiking | ho | 16:45 |
lisitsyn | wiking: so you have some troubles exposing the data? | 16:46 |
lisitsyn | I actually don't mind exposing whatever sgobject has | 16:46 |
@wiking | mmm | 16:46 |
@wiking | but you get it that here | 16:46 |
@wiking | we would be fucking | 16:46 |
@wiking | with the encapsulation pattern | 16:46 |
@wiking | right? | 16:46 |
@wiking | because you kind of force to expose *everythong | 16:46 |
@wiking | of sgo | 16:46 |
@wiking | anyparameter | 16:46 |
@wiking | etc etc | 16:46 |
@wiking | lisitsyn, i'll show u in a copy paste how super ugly this shit becomes | 17:05 |
lisitsyn | wiking: ok | 17:06 |
@wiking | no man this is even more shittier than you think :) | 17:07 |
@wiking | lisitsyn, https://pastebin.com/aDspYAaE | 17:10 |
@wiking | actually | 17:12 |
@wiking | this wont even work :)( | 17:12 |
@wiking | and note that this is not SAX but DOM | 17:13 |
@wiking | SAX reading woould be a suicide mission | 17:13 |
@wiking | but this doesn't let you do any thing like | 17:13 |
@wiking | serialize more than 1 object | 17:13 |
@wiking | into the file | 17:13 |
@wiking | :D | 17:13 |
@wiking | ok so forget it | 17:14 |
@wiking | this is even more and more worse | 17:14 |
@wiking | ok this wont work with dom | 17:14 |
@wiking | so we need sax | 17:14 |
@wiking | and if you need SAX | 17:14 |
@wiking | it's a mayham | 17:14 |
lisitsyn | wiking: yeah input is messy so far | 17:21 |
lisitsyn | output was easier indeed | 17:21 |
@wiking | lisitsyn, https://pastebin.com/4DUVTHmH | 17:22 |
@wiking | this should work imo | 17:22 |
lisitsyn | yes looks good enough | 17:23 |
lisitsyn | but not input | 17:23 |
@wiking | as well as the problem of encapuslation :P | 17:26 |
@wiking | note the story of SAX in reading | 17:28 |
@wiking | that's even more horrible | 17:28 |
@wiking | as you have to implement a full state machine :))) | 17:28 |
@wiking | this is just wasting LoC :) | 17:28 |
@wiking | that's my opinion | 17:28 |
@wiking | note that here we start to have the same story... as soon as we start to refactor the anything in the library | 17:29 |
@wiking | everything will break with serialization | 17:29 |
@wiking | because of the fact that we dont implement the serialization within the class itself | 17:29 |
@wiking | and this will need to be refactored everywhere | 17:30 |
@wiking | meaning in every serialization format | 17:30 |
@wiking | of course the other way round is getting shittier | 17:31 |
@wiking | as that means that you need to interface a serialization interface with primitives | 17:31 |
@wiking | .write(bool) .write(int) .write | 17:31 |
@wiking | which is getting ridicolous especially that the same shit has to be implemented in the visitor | 17:31 |
@wiking | tada | 17:31 |
@wiking | :) | 17:31 |
@wiking | btw | 17:36 |
@wiking | i'm pretty sure that with the current design | 17:36 |
@wiking | you just cannot do ser/deser :) | 17:36 |
@wiking | as this is the 10th iteration of this shit | 17:36 |
@wiking | and we are still the same point | 17:37 |
@wiking | i mean if we do it as an external case, i.e. your serialization fw is 'visiting' every method (so you need to handcraft everything for obj) then you have the current shitoooo situation that you see in the pasted code | 17:39 |
@wiking | namely when you start to read in recursively it kabooms | 17:39 |
lisitsyn | wiking: yeah but I don't really get what's exactly the problem in the design | 17:39 |
@wiking | now the other story ... namely that you let the objs drive serialization | 17:39 |
@wiking | you get where feature/cereal is being stuck | 17:39 |
@wiking | :) | 17:39 |
@wiking | so mumimu | 17:40 |
@wiking | this is asdf | 17:40 |
lisitsyn | wiking: who should drive the thing then? | 17:41 |
@wiking | there's a good reason | 17:41 |
@wiking | why everybody lets this specified by the class/obj | 17:41 |
@wiking | namely that you know yourself | 17:41 |
@wiking | what you need for getting back your state | 17:41 |
@wiking | just as in case of copy-ctor | 17:42 |
@wiking | btw the saddest thing of this all is that we are doing here something | 17:42 |
@wiking | that has been done by billions of times | 17:42 |
@wiking | :))) | 17:42 |
@wiking | http://rapidjson.org/md_doc_sax.html#Reader | 17:43 |
@wiking | look here what you need for the SAX parser | 17:43 |
@wiking | :) | 17:43 |
@wiking | and now there you need to implement a f-ing crazy FSM that you read in first the f-ing tuple for the f-ing sgobject to call the f-ing factory and once you have that you have to do the f-ing filling of the f-ing obj but then you arrive to the f-ing StartArray and you dont even know there the f-ing type of the f-ing array because then you still need to wait at least a next call from the input so that you kow the type of SGVector oooor SGMatrix you wanna | 17:46 |
@wiking | read in | 17:46 |
@wiking | man this is like loco | 17:46 |
@wiking | it's like 2002 again | 17:46 |
@wiking | where i parsed xml with libxml2 sax parser | 17:46 |
@wiking | and you know exactly | 17:47 |
@wiking | that this shit will somewhere just again blow up | 17:47 |
lisitsyn | ?\_(?)_/? | 17:47 |
@wiking | yeah exactly | 17:47 |
@wiking | i dont get it | 17:47 |
@wiking | why do we have to take this design | 17:47 |
@wiking | that makes the whole fucking shit | 17:47 |
@wiking | just like another piece of shit | 17:48 |
@wiking | :) | 17:48 |
lisitsyn | I don't really get what's the design in this context | 17:48 |
lisitsyn | key value? methods? | 17:48 |
@wiking | which case? | 17:48 |
lisitsyn | I mean you say design is broken | 17:48 |
lisitsyn | to unbroke we need what exactly is | 17:48 |
@wiking | yeah something in the design forces me to do things | 17:48 |
@wiking | that i cannot do | 17:48 |
@wiking | imo how we create objs | 17:49 |
@wiking | it's somewhere there | 17:49 |
lisitsyn | classifier('blabla') is bad? | 17:49 |
@wiking | nono | 17:49 |
lisitsyn | but plugins? | 17:49 |
lisitsyn | what is? | 17:49 |
@wiking | create("asdf", geenric_type) | 17:49 |
@wiking | because that assumes that you can inspect | 17:49 |
@wiking | or peek | 17:50 |
@wiking | the sax | 17:50 |
@wiking | because you need sax | 17:50 |
@wiking | as otherwise it's even more kaboom | 17:50 |
@wiking | :) | 17:50 |
lisitsyn | yes, sax | 17:50 |
@wiking | lisitsyn, https://www.nytimes.com/2018/04/19/opinion/standing-up-at-your-desk-could-make-you-smarter.html | 17:57 |
lisitsyn | wiking: doctor did not give me any prescriptions | 17:58 |
@wiking | show him this :) | 17:58 |
lisitsyn | she said like there is no such prescription per se | 17:58 |
@wiking | fucker | 17:58 |
@wiking | she just wanted more money :) | 17:58 |
lisitsyn | wiking: I asked my supply team what should I do to get a standing desk | 17:59 |
lisitsyn | :D | 17:59 |
@wiking | :) | 17:59 |
lisitsyn | wiking: apart from breaking my spine | 18:00 |
@wiking | noup | 18:02 |
@wiking | i thikn i already have a proof | 18:02 |
@wiking | that this cannnot be deserialized | 18:02 |
@wiking | :D | 18:02 |
@wiking | so i' | 18:11 |
@wiking | m out :) | 18:11 |
@wiking | working on something more fruitful :) | 18:11 |
@wiking | sorry | 18:11 |
@wiking | ttyl | 18:11 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4262 | 18:19 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4262 merged by iglesias | 18:19 |
@sukey | [https://github.com/shogun-toolbox/shogun] iglesias pushed 2 commits: | 18:19 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/b63e281ba37ca227d54036a924a445a47ed737e6 | 18:19 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/2102ffedefbcff7afe9ce98e28e4a90dd6ba8569 | 18:19 |
-!- iglesias [50fe9643@gateway/web/freenode/ip.80.254.150.67] has joined #shogun | 18:20 | |
iglesias | wiking: just checking out now that ny times article | 18:26 |
iglesias | they went all out with this one "The implication is that the more time you spend in a chair the worse it is for your brain health, resulting in possible impairment in learning and memory." wow | 18:27 |
-!- iglesias [50fe9643@gateway/web/freenode/ip.80.254.150.67] has quit [Quit: Page closed] | 18:35 | |
-shogun-buildbot:#shogun- Build deb1 - libshogun #393 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/10/builds/393 | 18:51 | |
-shogun-buildbot:#shogun- Build yakkety - libshogun #450 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/1/builds/450 | 19:30 | |
-!- travis-ci [~travis-ci@ec2-54-162-158-152.compute-1.amazonaws.com] has joined #shogun | 19:43 | |
travis-ci | it's Fernando J. Iglesias Garcia's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/369208959 | 19:43 |
-!- travis-ci [~travis-ci@ec2-54-162-158-152.compute-1.amazonaws.com] has left #shogun [] | 19:43 | |
-shogun-buildbot:#shogun- Build deb3 - interfaces #315 is complete: Success [build successful] - http://buildbot.shogun-toolbox.org:8080/#builders/37/builds/315 | 19:51 | |
-!- witness [uid10044@gateway/web/irccloud.com/x-euzmyvhpfazcmxpc] has quit [Quit: Connection closed for inactivity] | 21:37 | |
--- Log closed Sat Apr 21 00:00:33 2018 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!