rolling.classify: issue with training.corpus/test.corpus 'corpus cannot be used'

[meier-flo]

Hi,
I'm struggling with using an existing corpus as input parameters for rolling.classify(). After failing multiple times with my own data, I tried to simply use the Roman_de_la_Rose examples. However, loading the .txt files for the training and testing manually by using the load.corpus.and.parse() function did not work either (see image below). I'm pretty sure that the data format is the right one.

I looked at the code for the function on Github and stumbled over the following line:
if(is.list(parsed.corpus) == TRUE & length(parsed.corpus) > 1)

Maybe I'm just heavily mistaken, but shouldn't length(parsed.corpus) be only be >0 because otherwise, it would fail for single test_corpus?
Anyway would be nice to get some feedback on how to get it working, without having to transform every text into a .txt file and use the default approach (because that is working).

Thanks,
Florian