08:45 < jenkins-mlpack2> Project docker mlpack nightly build build #230: STILL UNSTABLE in 3 hr 30 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/230/

10:34 < ShikharJ> rcurtin: I started a page here: https://github.com/mlpack/mlpack/wiki/Testing-Guidelines-%28WIP%29 . I'll add more in due time.

13:46 < ayesdie> I think these testing guidelines will be great. I wanted to know how should we select the tolerance while testing the functions. I see 1e-5 being used as tolerance in most of the checks but some other tests have a tolerance of 1e-2.

13:51 < KimSangYeon-DGU> ShikharJ: Are you there??

13:53 < KimSangYeon-DGU> If you have a chance, please see https://github.com/mlpack/mlpack/pull/1713#discussion_r255323970. I want to know your opinion.

13:54 < zoq> ayesdie: The tolerance depends on the method, there is no single answer; some of the mlpack tests are probabilistic, so they may randomly fail, to reduce the number of failing test we sometimes have to increase the tolerance so that it works for a bunch of different random seeds. At the same time, we can't use a tolerance value that's too high, in which case we can't catch an actual issue.

13:54 < zoq> ayesdie: One example where we use a higher tolarance is the neural network code, mainly because we are restricted in the time we have to train the network.

13:56 < ShikharJ> KimSangYeon-DGU: I'll take a look soon.

13:56 < KimSangYeon-DGU> We can remove the unused parameter `visiblePenaltyGrad` by using directly `gradient` variable.

13:56 < KimSangYeon-DGU> Thanks!

14:02 < ShikharJ> KimSangYeon-DGU: Yeah, but the thing is, it is an alias variable, and not just any other variable. Ideally we can remove any alias in mlpack, but it won't help us with our readability right?

14:02 < KimSangYeon-DGU> Agreed

14:03 < KimSangYeon-DGU> I think the readability is important

14:04 < KimSangYeon-DGU> at the same time, removing the unnecessary things is important as well.

14:05 < KimSangYeon-DGU> So, I wanted to hear your opinion.

14:06 < ShikharJ> I don't really have any opinion on it, but if the static analysis check would fail because a warning would be considered as an error, then I don't mind removing it.

14:07 < ayesdie> zoq: thanks :)

14:08 < KimSangYeon-DGU> Judging from the opinions, I think it would be better to remove the `visiblePenaltyGrad` in this issue. I'll edit it soon.

14:10 < zoq> In some issues from the Static Code Check aren't really issues, in which case we can't whitelist those.

14:24 < KimSangYeon-DGU> Agreed. So, is it okay if I get the changes for passing the Static Code Check back?

14:24 < zoq> For the visiblePenaltyGrad fix?

14:25 < KimSangYeon-DGU> Both of visiblePenaltyGrad and model = nullptr

14:26 < zoq> Yeah, will whitelist them

14:26 < KimSangYeon-DGU> Actually, I think those are not a problem, although it fail to pass the static code analysis check

14:26 < KimSangYeon-DGU> Ah~ great.

14:26 < KimSangYeon-DGU> Actually, I was worried about it, while I eating meals.

14:28 < KimSangYeon-DGU> *having a meal. :)

14:28 < zoq> Have fun :)

18:36 < KimSangYeon-DGU> zoq: Thank you for resolving the issue. Now, the alias variable `visiblePenaltyGrad` fails to pass the Static Code Analysis checks. Would it be better to remove it??

18:37 < zoq> KimSangYeon-DGU: Just testing if the whitelist worked, will add the other issue as well :)

18:37 < KimSangYeon-DGU> Aha~ Ok!!

18:38 < KimSangYeon-DGU> Great. :)

18:40 < zoq> KimSangYeon-DGU: Okay, once this the build finished, we can merge the PR.

18:40 < KimSangYeon-DGU> zoq: Yeah!! :)

23:12 < rcurtin> KimSangYeon-DGU: thanks again for the awesome work with the diagonal GMMs. It is looking great

23:12 < rcurtin> (does Unicode go through IRC ok?) 👍

23:12 < rcurtin> (my terminal doesn't even display that, but it's supposed to be a thumbs up :))

23:13 < rcurtin> actually, apparently it must display ok because it ends up on the logs just fine :)

23:16 < KimSangYeon-DGU> rcurtin: Haha, the thumbs up looks small to me :)

23:17 < KimSangYeon-DGU> rcurtin: I'm very moved to your detailed reviews

23:19 < KimSangYeon-DGU> rcurtin: Thank you so much. :)

23:21 < KimSangYeon-DGU> The thumbs up is great :)

23:23 < KimSangYeon-DGU> 😊

23:34 < rcurtin> :)