fix(realtime): enable realtime publication for signature requests

[Jipperism]

• important note: subscribing to a table that doesn't have realtime publications enabled, disables existing subscriptions to all tables 🥴

Introduced in 70c8131. I think it would be good if we can find a way to double check at runtime whether all tables that are being subscribed to have realtime publications enabled.

[github-actions]

Coverage Report

Status	Category	Percentage	Covered / Total
🟢	Lines	25.35% (🎯 25%)	1097 / 4327
🟢	Statements	25.35% (🎯 25%)	1097 / 4327
🟢	Functions	61.05% (🎯 61%)	58 / 95
🟢	Branches	72.28% (🎯 72%)	180 / 249

File Coverage

No changed files found.

Generated in workflow #49 for commit 7d71aa4 by the Vitest Coverage Report Action

[bitbeckers]

@Jipperism to your point on checking for publications. Why not enable them for all tables?

https://supabase.com/docs/guides/database/replication#create-a-publication

[Jipperism]

@bitbeckers interesting idea. Not 100% sure if that would solve the problem, we'd have to verify if that is for all tables at the time the migration is run, or all tables from now on. If so, yeah why not?

[bitbeckers]

Looks like we can drop the publication and recreate it for all tables in a single migration: https://supabase.com/docs/guides/database/replication#recreate-a-publication

[Jipperism]

Yes, but that would then have to be added to every migration that creates a new table, right?

[bitbeckers]

Yes. Would a solution be to run a GHA after the migration that gets all entries from publication table and a list of all tables and finds the diff?

[Jipperism]

@bitbeckers is the idea that it then errors if not all tables are in the publication? Or it then creates a migration where we add all missing tables to the realtime publication?

Different approach: wrapper method for subscribing that checks the table being subscribed to against all tables in the realtime publication. That would solve the problem at startup time, so you'd see it before pushing. I think both would be fine, the wrapper method would keep the migrations cleaner.

[bitbeckers]

@Jipperism for the rest of the discussion, let's not get bikeshed here: #246

edit: sorry I though I already approved