Francis Nimick

@marklai1998 @patrick-yendou I dug out the solution I used to use before we switched off prisma entirely. Hope this helps? https://gist.github.com/fnimick/9fce65218a15b8b0ef3eb4592c20a0e4 I was tempted to try to make this a...

I believe the two-client approach does run the transactions / RLS code twice. The parent client opens a transaction, sets RLS, and then defers to a second extended client (either...

@marklai1998 out of curiosity what do you mean by the rls implementation varying? I assume different RLS strategies simply set different transaction variables, which we could easily support by passing...

> If we dont get this fixed in upcoming few releases we can think of making it a lib I'm not hopeful considering it took a full year for them...

Still happening with `0.32.1` with `postgres-js`. It also differs from the top level of the relational query to any included entities. ``` direct select of datetime [ { createdAt: '2024-08-09...

I've found the problem. The issue is the single query being generated for relational queries using an intermediate json representation, and converting a date to json converts to ISO. here's...

If you are calling `getUser` on requests and want to silence the warning as your session is guaranteed to be safe, see my approach in https://github.com/fnimick/sveltekit-supabase-auth-starter - I replace the...

Hm. If possible, could we eliminate the recursive fields and allow use of the form fields for well-defined fields? For my use case, I am using the non-recursive fields with...

Ah, I'm saying that if I have a recursive schema, but am using FormPathLeaves to access a non-recursive component, it still fails. This means that any recursive schema can't use...

@ciscoheat if you let typescript infer the type on the extended zod schema, it works - but that's because the recursive definition is now missing from the zod type! The...