move everything one directory up

[jakob-r]

http://mlr-org.github.io/mlr-tutorial/tutorial/ is just to redundant.
http://mlr-org.github.io/mlr-tutorial/release/html/ is how it should look like.
Any thoughts?
I would just need an ok and a lock to change all links.

[schiffner]

I'm not opposed, but would like to let @larskotthoff decide, because he built the directory structure.

[larskotthoff]

Fine with me.

[jakob-r]

I will do it now

[jakob-r]

Okay - i am mainly finished - waiting for travis to pass

[schiffner]

Big thanks!

[larskotthoff]

Ok, for some reason this fails in Travis (see https://travis-ci.org/mlr-org/mlr-tutorial/builds/69224059). I can't reproduce this locally.

The commits break the published tutorial, so I've reverted everything for now -- this needs to work while useR is going on. All the changes are in a new branch gh-pages-move that we can come back to next week.

[larskotthoff]

So the error seems to be completely unrelated -- Travis is and stays broken. I've pushed a hand-generated version so that the tutorial itself works at least.

I'm going to continue investigating what's happening next week (after useR) unless somebody has an idea what's going on here?

[schiffner]

Sorry, was away last evening, otherwise I could have clarified things earlier: @jakob-r 's changes are indeed unrelated. Before failing, the travis build #69 worked, but ran into a 50 min error. I restarted it several times, but had no luck. I will go over the tutorial next week and see if I can make things cheaper.

After installing mlr HEAD I can reproduce the error in building the tutorial locally.
It's related to the latest changes in plotThreshVsPerf(l. 98 - 102 in https://github.com/berndbischl/mlr/blob/c6e7d591cbdba2703971bd851fc62ead33845628/R/generateThreshVsPerf.R: use measure names for annotating the plots).
For the custom measure credit.costs calling get (in l. 99) causes the error. Apparently, when building the tutorial, credit.costs is not in the environment where get is looking.

Unfortunately, at the moment I can't reproduce the error itself.

[berndbischl]

Who wrote that code there? I mean using a "get" is like saying "Please R, hurt me, now".
Why is that needed?

[jakob-r]

Please discuss what is broken not because of the directory change in another thread.

[berndbischl]

Yes, we should keep issues separate. But I guess this might also be something rather for email. Let me just say one thing here, which I will repeat again in an email. Stuff like get() CANNOT be used, expect it is unavoidable. We need a proper programming style, somebody cannot write code ala "well it works on my machine, here, now".

[zmjones]

I think that maybe was the commit that lars merged because he thought it was only the 50 minute error? You should revert that commit though!

[larskotthoff]

Ok, all of that should be fixed now. I'll wait until after useR to reinstate Jakob's changes :)

[larskotthoff]

I will do this now.

[larskotthoff]

Done.

[schiffner]

Thanks, Lars.
I also did this:

9e7f1b2

mlr-org/mlr@86d6a4d

I hope we got everything now. :)

[larskotthoff]

Thanks!