no feedback on this in particular, but I love supabase. I use it for several projects and it's been great.
I was hesitant to use triggers and PG functions initially, but after I got my migrations sorted out, it's been pretty awesome.
I was hesitant to use triggers and PG functions initially, but after I got my migrations sorted out, it's been pretty awesome.
I think if there was a tightly integrated framework for managing the state of all of these various triggers, views, functions and sproc through source and integrating them into the normal SDLC it would be a more appealing sell for complex projects
You can then take it a step further but opting-in to use Branching [2] to better manage environments. We just opened up the Branching feature to everyone [3].
[1]: https://supabase.com/docs/guides/cli/local-development#datab... [2]: https://supabase.com/docs/guides/platform/branching [3]: https://supabase.com/blog/branching-publicly-available
Let’s say that we are using the Typescript sdk to make our app and need to do some fancy aggregation on a table that isn’t supported by Postgrest natively (specifically we can’t retrieve the data with Postgrest out of the box with its typical setup). Postgrest tells us that in this case we can do two things: create a view or make a Postgres function. Each has their pros and cons, but with either choice now we have this problem: some of our business logic is in sproc/function/view and some of it is in Typescript. In a typical setup using an ORM it would all be in Typescript.
The conventional wisdom is that db’s are dumb state holders and all of the business logic goes in the app - Supabase attempts to turn this on its head and say no actually it’s ok to store business logic in the db. But now if we do that we have a new problem: we don’t have a blessed path forward for where the line is on what goes where anymore. We don’t have good patterns for storing and managing this so other developers understand where to put things and how to work with our app anymore, because it no longer holds the principle of least astonishment. That’s what I mean by framework in this context.
Maybe all that is necessary here is a battle tested example project that demonstrates the “correct” way to make this demarcation. But as-is it steers me away from using Supabase for more complex projects if I even think they will need something that Postgrest won’t support without making a view or sproc/function