ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
travis-ci has joined #mlpack
< travis-ci> UtR491/ensmallen#13 (moead_exp - cb135ee : UtR491): The build was broken.
travis-ci has left #mlpack []
ImQ009 has joined #mlpack
ImQ009 has quit [Quit: Leaving]
< rcurtin> shrit[m]: I'm digging into the random forest serialization failure, but don't have a smoking gun yet
< rcurtin> I can see that for some reason the children of each decision tree node don't appear to be being deserialized correctly
< shrit[m]> rcurtin: Ok I see. At the same time we have no error that is coming from DecisionTree
< rcurtin> shrit[m]: I took a look at the output of a serialized random forest; it seems like the children of the decision tree are not being serialized
< rcurtin> currently digging into why...
< rcurtin> ha... I think I found it, but testing now
< rcurtin> the line in serialize() to serialize the children reads:
< rcurtin> CEREAL_VECTOR_POINTER(children);
< rcurtin> but... that doesn't do anything :-D
< rcurtin> my eyes just completely missed it the first ten times
< rcurtin> I think it just needs to be
< rcurtin> ar & CEREAL_VECTOR_POINTER(children);
< rcurtin> so, let's see if that's the case... I'll let you know when it compiles and I can run the tests
< shrit[m]> Oh, that is a silent killer error,
< rcurtin> yeah, not even a warning :)
< shrit[m]> It is passing right now, thanks
< rcurtin> oh, so you made the change and it worked for you?
< rcurtin> awesome
< shrit[m]> Yes
< rcurtin> ok... my automatic mlpack release script is getting pretty mature; I tagged 3.4.0 with the script and it looks right, now I'm waiting on the website to rebuild the documentation and I'll send an announcement email / make a release on github
< rcurtin> this was pretty easy so far, at least for this part of the release, and I think I can try to have mlpack-bot do it next time, just like ensmallen
< zoq> wow, nice
< rcurtin> those are mostly personal notes to myself, but ideally, there is nothing left on that list---I remove anything that's not automated as soon as it's automated
< rcurtin> a long time ago, the list had like 20 items and the last item was "drink", but now I've automated even that
< zoq> I remember the last step, how did you automate that step?
< rcurtin> so you can automate making the drink: https://www.youtube.com/watch?v=vcxn6E6yp0o
< rcurtin> and then you can also automate what to do with it: https://www.youtube.com/watch?v=2-mduHXTQy8
< rcurtin> it's not clear if the christmas music is necessary, but based on the video... it might be?
< zoq> haha this is great
< zoq> but don't you need ice?
< zoq> a lot of ice?
< rcurtin> true :)
< rcurtin> ok, mlpack 3.4.0 fully released
< rcurtin> this time it only took an hour or so, and most of that was waiting on the build script to build the markdown documentation
< rcurtin> so, *way* better than not so long ago when it took 4-6 hours of pretty manual labor
< zoq> hooray - super excited to see a new release
< rcurtin> yeah, and I will try to remember how easy this was!
< rcurtin> next time, my hope is to add a rel/release-mlpack.sh script, like ensmallen has, and set up the machinery in mlpack-bot to do the right thing when an automatically opened PR is merged