Yes so you can just use the storage module alone but it works a bit like a plugin system where you can add more plugins to the core module (probably best to read the wiki). But its not like there are extra dependencies because of that. Only dependencies of storage-kt and the core module which has dependencies storage-kt would need anyway

Looks great

Well done.

This question is so general / basic I wasn't sure if it belonged in #1006358244786196510 or not (happy to move it if so): I've been using Supabase (and love it!) for months now, but am increasingly unsure what the "point" of PostgREST is. i.e. I understand that at a high level it's an automatically created API for your DB, but my experience has been that I spend a lot of time just mentally translating the syntax from SQL to PostgREST, and invariably I end up trawling thru the PostgREST docs looking for fairly basic SQL features (i.e. group bys, etc.) that it doesn't seem to natively support. If there were an endpoint that simply accepted raw SQL queries and returned results as JSON, wouldn't that be much simpler / more powerful? I'm probably missing something, which is why I'm asking here. 🙂

Are triggers affected by row-level security?

I have found supabase to be great for trivial crud apps that do not have too much business logic - Saves an enormous amount of time and boilerplate. but once you go beyond the basics and start adding triggers to maintain business logic invariants - ie a closed task cant be edited and all tasks in project cant consume more resources than the project budget) - public views that hide tables in private schema - granular permissions (eg only project owner can mark a project as completed, while a person assigned to the project can update its status to blocked/pending) - in this case you cant use RLS and you have to add the logic to the triggers the verbosity of pl/pgsql really starts to show and it becomes a huge pain to maintain, incredibly tedious to test (using pgtap) and ends up as a big ball of mud. I acknowledge it could be just me, but so far i have not found an elegant and maintainable way to structure my code. is supabase/postgrest doomed to be used for trivial apps, or there is a way to - decouple the business logic into procedures that are easier to unit test - split the codebase into separate files? I have seen using \`\i ./task/triggers.sql \` inside a main

schema.sql

but I had hard time making it work because the dependencies can get nonlinear pretty quickly the DX just takes a nosedive after reaching (imo fairly low) amount of complexity, where it just becomes more comfortable to rewrite it using more traditional stack.

also naming things (triggers, rls, procedures) is hard! would be great if there was an opinionated framework:)))

i have exceeded my bandwidth usage of supabase to around 4 gb?

i am in my free tier, so will supabase block my account until i pay for it or what is going to happen?

Naming things is hard in general.

As I've mentioned in the thread for @jaypinho's question, you have multiple escape hatches, you can use a direct connection to your Postgres database and do all your business logic in a server side framework of your choice. You can also use edge functions to do all your business logic intensive part of your app and call those where necessary. I think what we are missing is a full featured open source app that makes use of edge functions, postgres functions, rls, triggers to show how one would go about building such an app.

> you can use a direction connection to your Postgres database and do all your business logic in a server side framework of your choice yeah, but then supabase becomes a classic hosted postgres provider and does not really add any value > You can also use edge functions hmm, this is interesting. So you would forbid updating the table in question by

client.from(..).update()

by revoking access and reimplement everything in the edge function?

woah I just realized Supabase raised $116m in total

It doesn't feel like a company that did 😄

To your second question, yes and no. Yes you can block `insert`/`update` using RLS but then you would have to use the service key inside of your edge function or access the database directly but you would need a way to validate the user at this point.

how can I sort a column with the Rest API Url?

I tried with: /rest/v1/Notes?sort=title:asc but it does'nt work

You asked this in help before here, that was the correct place. Please don't double post questions. Thanks

ok, sorry

can anyone clear this for me please?

If you're in free tier, you should not have to pay anything. If you exceeded the bandwidth, you probably have to upgrade your instance to a paid one (or wait 1 month later to have the usage reset if there is any)

its saying exceeded but i am able to use the storage without any issues

how do I get a user on the serevr side in nextjs ?

There is a #1006358244786196510 channel to ask this, please don't cross post.

tbh supabase should store the discord avatar when using the auth for it

How long does it usually take to set up a project on supabase?

Do you mean to get an instance up and running or to set up your project? Less than a minute for the first. No way to answer the second as it totally depends on your project.

this is weird, Google now acts like I'm trying to RECOVER my account when I use the Supabase authentication to log in

never did that before. would say "new login" maybe?

weird