Here I'll address some frequent questions I get asked when I introduce the project to Elixir developers.
TL;DR: for idiomatic code. And convenience.
I started to work on SwissSchema after noticing that using CRUD functions out of Phoenix generators brings a maintenance burden I don't want to have.
Ecto also lacks an idiomatic API to query Ecto repositories. Writing code blocks like Repo.get(User 123) is just a bit verbose, it often needs a case block, etc.
SwissSchema tries to be a one-stop-shop for this type of issues.
TL;DR: context modules should hold transactional functions, not CRUDs.
The functions that Phoenix generates when you use mix phx.gen.schema to generate code is also more code for you to maintain yourself. I don't think it's wise to maintain such basic functions (eg: tests, docs, etc.) in each project we build. It just doesn't make sense.
Also, I think that context modules should not contain CRUD code. Context modules are supposed to be the entry point for inner functionality (aka: a Facade). So, moving these CRUD functions to a shared FOSS project allows context modules to focus on what they're supposed to do.
Some functions in Ecto APIs are not tied to a schema, they accept queries instead (or something else that's not directly tied to a schema module). Implementing these functions might cause confusion, since their interfaces would differ from their Ecto equivalents (SwissSchema functions are always tied to a schema). It might even get developers confused by the API differences (eg: the Ecto one expects a query, but the SwissSchema one expects an ID, of a keyword list of params), so these are not getting equivalents in SwissSchema (at least not with the same names):
Below is a list of functions whose absence is enforced:
exists?/2(Ecto.Repo)
It's the same license used by Elixir.