When during the operation an error occures, the whole (pending) operations list is cleared. This is annoying and unneeded, because if there is an actual mistake with the choice of operations it can still be cleared by hand. However when only a small mistake triggered the error (I had a mount point open), one has to redo all the steps, although a simple unmount fixes the error. Therefore I propose to keep the list of pending operations after an error occured - on a succesfull run the list should still be cleared.
I don't think there is any easy way to do that. Please suggest if there is. After the error happens we must rescan existing partitions, otherwise data corruption might happen. And rescaning involves removing all operations from operations stack, the same happens even if you refresh before applying operations.