00:11 < naywhayare> I have to step out for dinner; I'll be back...

00:40 < george_> sorry, didn't see these messages till now

00:42 < george_> gcc 4.4.7

01:45 < naywhayare> hmm, that's a pretty old gcc version... RHEL6 I assume

01:45 < naywhayare> (or scientific linux / centos)

01:45 < naywhayare> I know that the Armadillo maintainer doesn't have much sympathy for old compilers especially because Armadillo often breaks compilers...

01:46 < naywhayare> my off-the-cuff advice would be "try with an older version of Armadillo", maybe 3.xxx or something; if that doesn't help, I'm happy to poke through a full list of errors to try and understand what's gone wrong

01:55 < george_> interesting suggestion, that might work. I had not independently tested the armadillo install, but thought it might be a good idea. I'll post the errors in case any of it is helpful.

01:56 < george_> is it bad form to post 75 lines in IRC?

02:01 < naywhayare> freenode might kick you if you pasted them too fast

02:01 < naywhayare> pastebin may be a better solution

02:09 < george_> http://pastebin.com/Sr8148SF

02:15 < naywhayare> ok, this is getting into C++ horror land

02:15 < naywhayare> a similar problem is kindly documented by one Dave Steffen here: https://svn.boost.org/trac/boost/ticket/6362

02:16 < naywhayare> my understanding is this: anonymous enum types are allowed now, but when gcc 4.4 was released, this part of the standard was either incomplete or different than it is now (though I don't quite follow; I don't think this is a C++11 feature? I'm not sure)

02:17 < naywhayare> this implies that on non-gcc-4.4, you probably won't encounter this problem, but what I'm trying to figure out is what is raising the issue in the first place

02:17 < naywhayare> the actual error doesn't have to do with Armadillo but instead with std::__traitor<std::__is_integer<unsigned int>, std::__is_floating<unsigned int> >::<anonymous enum> >::value

02:17 < naywhayare> which isn't Armadillo... it's the standard library

02:18 < george_> ah

02:18 < naywhayare> this makes me wonder if something is really, really weird on the system you are working on, like gcc using a libstdc++ that was released after the gcc version (can you even do that? I don't know why or how)

02:19 < naywhayare> I would think that finding a newer compiler would alleviate the issue, but if you want to dig deeper into this particular nightmare I am more than happy to help because these kinds of oddities interest me :)

02:21 < naywhayare> in my own experiences on RHEL6 with gcc 4.4.7, mlpack and Armadillo both compile fine, but in my experience cluster setups tend to be... special...

02:26 < george_> am on Skype call, sorry for lack of responsiveness

02:26 < naywhayare> not a problem -- I'll be here probably until past midnight

02:26 < naywhayare> (EST)

04:54 < naywhayare> looks like the Armadillo warning message was fixed in r17250 about six weeks ago: http://www.mlpack.org/trac/changeset/17250

05:20 < george_> This is interesting, the cluster I am using has quite old software, in fact installing anything new is a pain. I can probably wrangle enough disk space to install everything I would need from scratch, though the system also has an Intel compiler which I could try.

05:22 < george_> I am more than happy to receive help with this, as it is quite important to a project I am working on

05:22 < george_> so if you are interested I'll happily cooperate.

05:53 < naywhayare> george_: try the intel compiler first... that's probably the path of least resistance

05:53 < naywhayare> newer gcc can be installed by hand locally, but it's tedious and sort of a pain (frustration level: 4/10)

17:14 < tarun_ml_enthusi> finally got mlpack setup and working :D

17:27 < tarun_ml_enthusi> i had signed up for the mailing list a few days back but havent recieved any mails...is there a problem with my subscription or are the discussions not publically viewable or something like that?

17:57 < tarun_ml_enthusi> to get started off as a beginner...could you show me some easy bugs to try and fix? i couldnt find a page where the bugs are classified by difficulty level

18:21 < naywhayare> tarun_ml_enthusi: the mailing list is pretty low-traffic; no messages have been sent there recently

18:22 < naywhayare> the mlpack-svn mailing list is higher traffic, but it's just commits to the repo so maybe it's not so interesting...

18:23 < tarun_ml_enthusi> oh okay...can you help me getting started with bug fixing? i havent used trac before ...where can i find some bugs which can be solved by a beginner? and how do i upload changes?

18:23 < tarun_ml_enthusi> is there a github repository where i can push changes or something?

18:23 < naywhayare> the tickets on Trac aren't classified by difficulty because it's often hard to know which ones will be difficult to some people and which will be difficult to others...

18:23 < naywhayare> you can just attach a patch (which you can create with 'svn diff') to the ticket, if you find one that you find interesting

18:25 < naywhayare> I should note that many of the tickets on Trac are specific to a certain algorithm, so it might be best to look for a ticket that concerns an algorithm you know something about

18:25 < tarun_ml_enthusi> svn diff...okay ill look into that..thanks

18:26 < tarun_ml_enthusi> yeah your right

18:26 < tarun_ml_enthusi> ill look into bugs for linear regression and k means

18:27 < naywhayare> another idea is to look through the implementation for those two algorithms and see if you have ideas on how to improve them (i.e. make them more flexible, or make them faster)

18:29 < tarun_ml_enthusi> like trying out different data structures and comparing the results right?

18:29 < naywhayare> or, implement an entirely new algorithm, if you see something that mlpack is missing

18:29 < naywhayare> yeah, that's a possibility

18:29 < naywhayare> anyway, I have to run... I'll be back on Monday

18:29 < tarun_ml_enthusi> ok ..thank you :)

18:29 < tarun_ml_enthusi> ill try my best to contribute :D

18:30 < naywhayare> :)

18:30 < naywhayare> talk to you later