[Feature] Idempotence - Generate random dataset deterministically

[bhagenbourger]

I propose to add a feature to generate the same output several times.

To add deterministic way, I suggest to add a seed and generate all values from this seed. For the same seed set as parameter of fakelake, the output will be the same.

For example : fakelake generate --seed xxxxxx path/to/schema.yaml If seed is not passed as parameter, a random seed is generated. After file generation, the used seed is printed.

This feature enables to generate the same dataset in different formats (CSV and PARQUET for example). Also, it easier to share a seed than a full dataset if you want to reproduce something on another environment.

[vianneybacoup]

The only concern I have for now is that Parquet is generated with threads. So either:

• we don't care about order when given a seed (not fan of this but why not)
• we control the order it is written in the file
• we have to remove the parallelisation when provided a seed

[bhagenbourger]

Thank you for these informations, will think about it.