ABSTRACT
When working on a codebase, whether it is a decades-old linear algebra package or just a small experiment script, it is important to be able to work without worrying about losing data.
One approach to preventing data loss is to save a copy of your files with a different name each time, usually appended with the date or revision number. Invariably, this ends up with a directory littered with files with names like “sample_data_v2,” “preprocessor_ final_rev_2,” and “paper_draft_v2_2015_10_21_FINAL_3b.” This can make finding the most authoritative version of a file difficult.
