All models in the app are using UUID (generated by postgres: gen_random_uuid() from pgcrypto extension) instead of traditional sequential ids.
Benefits:
- for distributed systems gain ability to assign primary key values without collisions
- unable to guess records just by incrementing/decrementing ids or to get some insights on how big is the dataset
- prevents from typos as UUID needs to be the exact value
Dark side:
- hard to communicate them without copy/paste - don't try to say them
- rows are sorted randomly so without created_at you can't tell the order
- more memory cause it's bigger than just integers
Setup: https://github.com/dmp0x7c5/travel-blog/pull/4/files