hackalog / easydata

A flexible template for doing reproducible data science in Python.
MIT License
108 stars 22 forks source link

Build rpathlib and use it to replace pathlib #229

Open acwooding opened 3 years ago

acwooding commented 3 years ago

We love pathlib. It is great for using paths in a platform independent way (windows/mac/linux). We use it in easydata when we give everything in an easydata repo a standard reference location relative to the project base path. This works great when everything is local. But breaks when you want to start to use the cloud. We started using fsspec for this. We now have a bunch of if local: use pathlib. If remote: use fsspec in the code. What we'd like, is to unite these two with a common API so we can continue to use paths, like in pathlib, but magically have it work even via Azure or AWS. Hence, rpathlib. Here's the spec in picture format: Untitled_Artwork 1