--- Log opened Sun Oct 02 00:00:52 2011 | ||
blackburn1 | oh I was wrong with that extension | 00:46 |
---|---|---|
-!- blackburn1 [~blackburn@188.168.4.116] has quit [Quit: Leaving.] | 01:42 | |
-!- blackburn [~blackburn@188.168.3.99] has joined #shogun | 10:19 | |
-!- blackburn [~blackburn@188.168.3.99] has quit [Ping timeout: 260 seconds] | 17:25 | |
-!- blackburn [~blackburn@188.168.14.11] has joined #shogun | 19:20 | |
@sonney2k | blackburn, which extension? | 19:24 |
blackburn | sonney2k: I still trying to get my kernel LTSA work right | 19:24 |
@sonney2k | ok | 19:25 |
blackburn | sonney2k: we can't disable pthreads? | 19:30 |
@sonney2k | ? | 19:30 |
blackburn | sonney2k: no --disable-pthreads option | 19:31 |
@sonney2k | I see | 19:31 |
@sonney2k | indeed... | 19:31 |
blackburn | sonney2k: have you seen a cool feature I added? now possible to set target dimensionality to -1 and it will be reduced by 1 | 19:34 |
@sonney2k | isn't hat confusing? | 19:35 |
blackburn | I don't know, is it? | 19:36 |
@sonney2k | well you need to document it at least | 20:02 |
blackburn | oh | 20:14 |
CIA-3 | shogun: Sergey Lisitsyn master * r246ccf6 / src/shogun/preprocessor/DimensionReductionPreprocessor.cpp : Fixed wrong destruction of DR preprocessor base class - http://git.io/STWLGA | 20:15 |
CIA-3 | shogun: Sergey Lisitsyn master * r51eeb3d / (7 files): Improved multithreading codestyle, fixes some structure - http://git.io/xdq67Q | 20:15 |
blackburn | sonney2k: last one commit reduced LoC related to multithreading VERY much | 20:16 |
@sonney2k | blackburn, so it is not using that weird syntax but just { } are enough? | 20:17 |
@sonney2k | blackburn, the only problem I see is that it is unreadable - I mean you have like 20 args | 20:18 |
@sonney2k | and it is totally unclear which one is for what | 20:18 |
blackburn | sonney2k: not more unreadable than it was :D | 20:19 |
@sonney2k | well before it at least said start_idx=0 | 20:19 |
@sonney2k | now it says 0 somewhere hidden among those args... | 20:19 |
blackburn | sure, but INFINITY lines of .x=y is crazy too | 20:19 |
@sonney2k | no | 20:19 |
@sonney2k | it helps to understand what you do and you could do that on the same line too | 20:20 |
blackburn | it is just parameters passing, I think it is ok | 20:20 |
blackburn | not the main readability problem :) | 20:20 |
blackburn | e.g. DSYEVR routine have 22 args and it is not that difficult to read | 20:21 |
blackburn | for me at least | 20:21 |
@sonney2k | blackburn, if there is a bug in creating this structure there it is very difficult to figure out which arg is the right one | 20:21 |
@sonney2k | I consider any function with more than a handful of args 'broken by design' | 20:21 |
blackburn | sonney2k: are you suggesting to revert the commit above? :) | 20:22 |
blackburn | I spent a hour doing that :D | 20:23 |
@sonney2k | I am suggesting to add the .x annotation | 20:23 |
@sonney2k | otherwise reverting is better yes | 20:23 |
blackburn | sonney2k: it is supported only by C99 standard | 20:23 |
@sonney2k | before I had a chance to understand it - now I don't | 20:23 |
blackburn | sonney2k: I will send a bomb to berlin :) | 20:25 |
@sonney2k | we have enough terrorist :) | 20:25 |
blackburn | spent too much time on that | 20:25 |
@sonney2k | one more doesnt' make a difference | 20:26 |
blackburn | terrorist? do you have any? | 20:26 |
@sonney2k | why not use the C99 syntax? | 20:26 |
blackburn | sonney2k: produce warnings | 20:26 |
@sonney2k | argh | 20:28 |
@sonney2k | why is C99 stuff not in C++ ?!! | 20:29 |
blackburn | sonney2k: that is the problem, yeah | 20:29 |
@sonney2k | then we cannot use it :( | 20:30 |
@sonney2k | hmmhh, maybe we should rethink the whole threading strategy | 20:30 |
@sonney2k | I mean wouldn't it be possible to create some class for this business | 20:31 |
@sonney2k | (templated I guess) | 20:31 |
@sonney2k | and then add the usual functionality in there like the start/end things | 20:31 |
blackburn | sonney2k: well yes, all of them contain some index start,step,stop | 20:31 |
@sonney2k | I mean we have this appearing 20 or more times now | 20:32 |
@sonney2k | it is getting #W$$!?!#?@ | 20:32 |
blackburn | sonney2k: a LOT of work | 20:33 |
@sonney2k | also we always solve the start the threads / join the threads / deal with error handling | 20:33 |
@sonney2k | but wisely spent | 20:33 |
@sonney2k | my suggestion would be to have start, end always in there and some templated thing called user_data or so | 20:35 |
@sonney2k | that contains the struct with required metadata for the task | 20:35 |
blackburn | sonney2k: how to revert commit? | 20:36 |
@sonney2k | git revert <id> | 20:36 |
@sonney2k | then push again | 20:37 |
CIA-3 | shogun: Sergey Lisitsyn master * r15ebd97 / (7 files): | 20:38 |
CIA-3 | shogun: Revert "Improved multithreading codestyle, fixes some structure" | 20:38 |
CIA-3 | shogun: This reverts commit 51eeb3d7aeaebeb41297d6db822a073d03679e6b. - http://git.io/-pXtNQ | 20:38 |
blackburn | sonney2k: yes, it could be one more bug-generating thing | 20:38 |
blackburn | I would better stay it as it was | 20:38 |
@sonney2k | blackburn, I guess you should open a trac ticket for this | 20:39 |
blackburn | sonney2k: for what? | 20:39 |
@sonney2k | for this threading class | 20:39 |
@sonney2k | I guess it will need a couple of iterations to be worth anything | 20:42 |
blackburn | , | 20:43 |
blackburn | ? | 20:43 |
@sonney2k | if we have to write 50 LoC to use that class it is not really an improvement | 20:43 |
@sonney2k | alright - I will torture myself today with installing windows 7 again | 20:44 |
@sonney2k | I hope I manage to do this somehow | 20:45 |
@sonney2k | I never ever considered that this could be impossible when I have 2 harddrives... | 20:45 |
blackburn | hah | 20:45 |
blackburn | it would be impossible to install win7 to the second partition | 20:46 |
blackburn | I Imagine what is the problems with multiple anything | 20:46 |
@sonney2k | blackburn, well osx installed w/o problems to any harddrive | 20:47 |
@sonney2k | even external usb drives | 20:47 |
@sonney2k | and linux same thing | 20:47 |
blackburn | no way to install win7 to flash drive :D | 20:47 |
@sonney2k | the problem is that I have to open up my notebook | 20:48 |
@sonney2k | remove one harddisk | 20:48 |
@sonney2k | and then install on the other | 20:48 |
@sonney2k | and then pray that it works when I connect both of them | 20:48 |
blackburn | sonney2k: does you notebook have two slots? | 20:49 |
@sonney2k | blackburn, I removed the dvd drive | 20:49 |
blackburn | crazy | 20:49 |
@sonney2k | and connected a sata drive there | 20:49 |
@sonney2k | well when you have kids you suddenly have too many photos/movies | 20:49 |
@sonney2k | and of course I was hoping to get shogun to run on cygwin / osx | 20:50 |
CIA-3 | shogun: Sergey Lisitsyn master * r9698ed3 / (4 files): Some necessary improvements from the reverted commit - http://git.io/eKzzgQ | 20:50 |
blackburn | hah | 20:50 |
@sonney2k | alright - if I won't be online in a few hrs - windows 7 killed my linux install | 20:52 |
@sonney2k | cu | 20:52 |
blackburn | okay see you | 20:53 |
-!- Guest56770 [~sonne@f053040237.adsl.alicedsl.de] has joined #shogun | 21:16 | |
-!- Guest56770 [~sonne@f053040237.adsl.alicedsl.de] has quit [Ping timeout: 248 seconds] | 22:33 | |
-!- blackburn [~blackburn@188.168.14.11] has quit [Ping timeout: 252 seconds] | 22:43 | |
-!- blackburn [~blackburn@188.168.5.143] has joined #shogun | 22:45 | |
--- Log closed Mon Oct 03 00:00:57 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!