Implementing any kind of machine learning requires some kind of persistent memory for the algorithm, but the only way to get some output from the code is to use ‘say’ thingy or trying to hover some global vars in a debug mode. Both of them are truncated and not available for selecting and copying, so this way is extremely painful.
We are going to try transitioning the hover debugger (which should start working much better) to a pinnable Treema popover, which should not only let you see all the state and drill down as needed, but also activate the Treema’s JSON copy feature to grab all the data you need. When that’s in, let me know how it goes.
It will be better, but in this case we will still need to copy-paste data manually.