@madebyfabian@kmcnam1@supabase It does block both UPDATE/DELETE without WHERE on Supabase. Do note however that this is not an inherent feature of PostgREST but an integration with https://t.co/HJgXhq87bW
@DanielLockyer Notably the REST API has an additional process: JWT authentication (which gives you a trusted context for RLS, a direct connection doesn't). The newest release has a JWT cache which reduces this process time. Try upgrading, see https://t.co/nqpMCIcn44
@martindonadieu@kiwicopple@PostgRest Hey! v14 is still not in prod as it requires extra testing to be integrated on the Supabase platform. Please check https://t.co/F23cOf0xb9 for updates.
@audreyt@zacstewart@TAIWANmoda FYI. This wasn't enough at all in my case, despite having more than 8 years of github experience. They sent my application back.
@machinegenie@supabase A table column like "my.dotted.col" will also need to be quoted on the URL when sending it to PostgREST. Likely something that needs to be addressed on supabase-py. For more details, see: https://t.co/feaIjwMhws
@AutisticOvrflow@parzerp@kuberdenis On regular PostgreSQL installation projects, you won't have the above problem since you have to explicitly GRANT privileges on the `public` schema db objects. Pretty much following https://t.co/BXvb0XBi6h to the T.
@AutisticOvrflow@parzerp@kuberdenis > PostgREST thing that just gives public access to your public schema.
The above is false, it doesn't. PostgREST just follows the privileges of the database roles, which on Supabase are "open" by default for ease of use but you can disable this, see https://t.co/lhKHi0GpJa.
@tristanbob@dshukertjr@supabase HTTPS connections are already enabled for PostgREST, this is done by the proxy (Kong on Supabase). https://t.co/C2tc8Guii5
@jscheel@kiwicopple Yes, it doesn't need a restart. See:
> To manually reload the cache without restarting the PostgREST server..
> NOTIFY pgrst, 'reload schema'
>https://t.co/9D1zb33WFC
@yrashk I've seen other rdbmses support a CREATE OR REPLACE TABLE for this purpose. Perhaps that plus some type of "migration constraints" could solve the problem? I also believe migrations need a lot of improvement.
@zombodb > sql function to do it, but those are a bit more complex to define since you have to repeat the return columns.
I've stumbled into that before. I thought it would be great if PostgreSQL supported return type inference like `CREATE FUNCTION... RETURNS AUTO`.
My first @PostgreSQL feature just got committed!
`pg_basetype` allows obtaining the base type of a domain.
Just a convenience function to avoid a more involved recursive query.
https://t.co/2P8Q7UbkGN