14:11 < zoq> rcurtin: I think #972 is ready to be merged, let me know if you have any thoughts

14:46 < Trion> zoq: I was getting "couldn't find erlport.cli" so I remaked it, now getting "python :"builtins.SyntaxError", '(unicode error) 'unicodeescape' codec can't decode bytes in position 289-290" in server when agent creates an environment

14:57 < rcurtin> zoq: I agree, I think it looks fine, maybe the bibtex URL for the adam paper cna be added after merge

14:57 < rcurtin> easy change :)

14:59 < zoq> Trion: Have you modified any file?

14:59 < zoq> rcurtin: Yeah, easy change

14:59 < zoq> Trion: Maybe you can start with a clean project?

15:02 < Trion> Nope, files were same, all modifications were after the couldn't find erlport.cli :P

15:03 < zoq> And with "I remaked it" you mean calling 'mix deps.get' right?

15:04 < Trion> nope, "make release" in deps/erlport :P

15:04 < zoq> rcurtin: "non-human contributor" :)

15:05 < zoq> Trion: Can you test if it works with 'mix deps.get'?

15:05 < Trion> mix deps.get says everything is upto-date

15:09 < rcurtin> yeah, I think we are still many steps away from having a "real" non-human contributor, but it seems like maybe CodeAi could be a useful tool

15:09 < rcurtin> we'll see what else it comes up with

15:15 < zoq> Trion: hm, so if you haven't modified the worker.py file, can you remove line 288-291 (envs.monitor_start(instance_id, directory, force, resume) etc.) and try again, I'm not sure what the error could be ...

15:16 < zoq> rcurtin: Yes, I wonder if they found other issues.

15:18 < Trion> Still didn't work.. I'll try with a clean copy then

15:20 < zoq> Trion: Okay, let me know if that helps

15:29 < Trion> zoq: "No module named erlport.cli" is back

15:37 < zoq> Trion: I think you already tested 'pip uninstall erlport' as suggested in: https://github.com/hdima/erlport/issues/32

15:39 < Trion> Yeah erlport doesn't exist, so uninstall doesn't work

15:40 < zoq> Trion: so make release resolves the first problem but results in another right?

15:41 < Trion> make release in erlport folder gets server to run :P but then it doesn't work for environmetn

15:42 < zoq> What's your OS and python version?

15:43 < Trion> Ubuntu gnome and python 3.5

15:43 < zoq> Ubuntu 14.04?

15:43 < Trion> nope 16.04

15:47 < zoq> never tested the code with python3.5

15:48 < Trion> It used to work few weeks ago, today it suddenly became "nah.."

15:49 < zoq> Can you test if the example code works? https://github.com/hdima/erlport

15:49 < zoq> I can try to replicate your setup later today

15:55 < Trion> Yeah the example command "{ok, P} = python:start()" is also giving Syntax error "keyword argument must be followed by space after: python:"

16:04 < zoq> I guess, what you could do is to use python2.7 instead of python3.5

16:05 < rcurtin> delivery of masterblaster to the new location should be within 90 minutes according to UPS

16:06 < rcurtin> I updated DNS too, so hopefully the folks in Springfield will get it racked and plugged in quickly :)

16:06 < Trion> What is masterblaster? :P

16:06 < zoq> cool

16:06 < rcurtin> Trion: it's the big build server that Symantec got for mlpack when I joined the company

16:06 < rcurtin> it runs the Jenkins instance we use for benchmarking, matrix build, testing, etc.

16:07 < rcurtin> I would send a link but it is currently offline :)

16:07 < Trion> Ooh nice!

16:11 < zoq> Trion: Another idea is not to use the latest erlport version.

16:12 < Trion> mix deps.get doesn't get the latest?

16:12 < zoq> Trion: change line 34 in mix.exes to {:erlport, git: "https://github.com/hdima/erlport.git", ref: "50b16d7ffecc1cadd367e68de723eea0580077c2"},

16:12 < zoq> Trion: it does

16:13 < Trion> hmm maybe it doesn't install it as python dependency

16:13 < zoq> Trion: maybe the latest change doesn't work for you

16:18 < Trion> nope :-( still "unicode error"

16:20 < Trion> It says 289-290 in 38 in worker.py is invalid

16:20 < zoq> okay, I'll see if I can reproduce the issue later today, until then feel free to use the demo server

16:20 < zoq> you can try to delete the lines

16:21 < Trion> ok

16:50 < systematic> rcurtin: I think you hit it right about the chatbox. That is a big problem with machine learning today. Namely, you can't teach the computer something that you do not yourself know.

16:53 < Trion> :P Deleted the comment at that line, then JSON decoder error poped up, I hope peace can be established with erlport

16:54 < zoq> Trion: ah right, if you delete the lines you also have to comment "env.monitor.start("./dummy/", true, true);" in example.cpp

16:55 < systematic> I think that the best results would come from giving the computer raw data rather than trying to get the data from users.

16:55 < Trion> then I won't have a recording of my funky agent, right?

16:56 < zoq> Trion: yeah, but if that works, we can probably find a way around the issue.

16:57 < Trion> systematic: That is possible but the amount of depth in that case would need lots of firepower :P which many tech giants will have hard time affording

17:03 < Trion> zoq: JSON decoder error again.. erlport seems to have some severe issues

17:06 < Trion> have to go for now, I'll check logs, try the setup in free time :P I hope it is easier to replicate

17:11 < systematic> trion: So, option B, have experts supply data.

17:25 < systematic> *option C

17:27 < zoq> Trion: What's the exact error you get if you use: https://gist.github.com/zoq/c3e4f90740399c6d42ac9e6f66a2c53c

20:22 < rcurtin> not sure why gitdub is going crazy and sending way too many commit emails

20:24 < zoq> rcurtin: it's sending out of the blue?

20:26 < rcurtin> after I pushed to mlpack-2.2.x

20:27 < rcurtin> it seems like maybe gitdub was not aware of those branches, but that doesn't seem right because it has sent emails for those branches in the past