Add more context to unit tests and best practices

[mirnawong1]

Contributions

• I have read the contribution docs, and understand what's expected of me.

Link to the page on docs.getdbt.com requiring updates

Per slack convo, the following unit test best practices/recommendations from @graciegoheen would be beneficial to add to the docs so users understand what works best for them:

Inline vs. fixtures

• inline is better when you have a small number of edge cases, and you want to be able to see that data in your yml file that defines your unit test
  • i think it’s easier to troubleshoot a unit test when everything is in the same file
• fixtures are better when you have lots of edge cases, so if you put it all in one yml file the file would be GIANT
  • i also think fixures are better for the sql format
  • but then you have to jump around between the fixture file and the yml file

dict vs. csv vs. sql

• dict is the default, it’s always inline so contains your data in the yml file for easy viewing
• csv is what a lot of folks are used to if they’ve been using a unit testing package (via seeds) - so this is recommended for folks that are migrating from unit testing package, or who have less technical folks that prefer looking at csv
• sql is for expanding the type of data you can unit test
  • can unit test more data types (think structs, geography, etc.)
  • total control over the sql
  • can unit test models that depend on ephemeral models
  • BUT when using this format you must define mock data for all columns

What part(s) of the page would you like to see updated?

https://docs.getdbt.com/docs/build/unit-tests

Additional information

No response