01:25 < rcurtin> mikeling: have you had a chance to add the test for the weighted decision tree learning that I suggested?

01:25 < rcurtin> I was hoping to look into that tomorrow, so if you haven't, no worries, I just wanted to avoid duplicating work if possible :)

10:10 < zoq> If we set the matlab search path: https://mathworks.com/help/matlab/matlab_env/what-is-the-matlab-search-path.html the tests should pass.

11:32 < zoq> hm, is there something wrong with the list?

11:36 < mikeling> rcurtin: Oh hey! I miss the message somehow. Please take it if you need, thank you!

11:36 < mikeling> :)

11:59 < jenkins-mlpack> Project mlpack - git commit test build #2725: FAILURE in 23 min: http://masterblaster.mlpack.org/job/mlpack%20-%20git%20commit%20test/2725/

11:59 < jenkins-mlpack> * Marcus Edel: Independently initialize each layer of a specific network; this could

11:59 < jenkins-mlpack> * Marcus Edel: Add test for the network initialization method (test with each

11:59 < jenkins-mlpack> * Marcus Edel: Remove unused parameter and simplify the cell update step.

11:59 < zoq> working on it

12:59 < rcurtin> zoq: did you really push that many commits, or is gitdub acting strangely again?

13:01 < zoq> rcurtin: I made 3 commits

13:02 < zoq> to master and othing to the mlpack-2.2.x branch

13:02 < rcurtin> ok, I see... :)

13:08 < zoq> Okay, the next time, I'll should test the code with clang and gcc.

13:24 < rcurtin> no worries, your success rate with commits is still much higher than mine :)

13:42 < zoq> nice now the windows build failed with a nuget timeout error

13:49 < rcurtin> sigh :(

13:54 < jenkins-mlpack> Yippee, build fixed!

13:54 < jenkins-mlpack> Project mlpack - git commit test build #2726: FIXED in 42 min: http://masterblaster.mlpack.org/job/mlpack%20-%20git%20commit%20test/2726/

13:54 < jenkins-mlpack> Marcus Edel: Define the model for each init rule to fix gcc issue.

14:18 < rcurtin> wow, that's really fast, it took -2 seconds to run...

14:24 < zoq> yeah :) I forgot to map the error for the IRC message ...

14:24 < rcurtin> I guess, for the matlab search path, I would have to add every methods/*/ directory, does that seem right to you?

14:27 < zoq> yeah, export MATLABPATH=$(shell pwd)/methods/matlab/ should do

14:31 < rcurtin> oh, right

14:31 < rcurtin> I should have checked...

14:31 < rcurtin> for some reason I thought it was laid out like mlpack where each method had its own directory :)

14:39 < rcurtin> now we can wait for a long time to see if I fixed the issues, since the checkout job now takes a couple hours because it builds all the libraries

14:39 < rcurtin> I guess, I could do the build step in parallel

14:40 < zoq> yeah, I guess you could run it in parallel

14:41 < rcurtin> yeah, I am changing the scripts now

14:41 < zoq> but I don't mind to wait :)

15:14 < rcurtin> ok, that should work better now, we can save some time :)

15:20 < zoq> right, just use all cores to build the libs

20:18 < rcurtin> weird, a segmentation fault in the python xml unittest runner...

20:31 < zoq> yeah, weird haven't found a solution yet, just a "workaround"

20:34 < rcurtin> oh, do you have a workaround for it?

20:34 < rcurtin> I was still trying to narrow down the cause

20:35 < rcurtin> it seems like the test XMLs are generated anyway, but maybe there is a problem with one of them, I haven't checked fully

20:39 < zoq> workaround is probably not the right term, since jenkins fails I used 'make checks || mkdir -p test' that way the build passes and the status depends on the number of failed tests, definitely not the best solution

20:39 < rcurtin> I think that's an ok workaround for now

20:39 < rcurtin> I'll go ahead and add that in