Datasink nodes don’t work on numpy arrays as nodeoutputs. I could just the pickle the output within the node, but I need to somehow give context in order to save the file with a unique name. I thought datasink could help with that.
eg. the working directory location of the result file would be fine if I could move that result file and its higher directories to someplace other than a scratch directory.
That would be a great and very well appreciated addition! TSV/CSV file export would also be great.