Ponder

Should Ponder replace SQLite with PGlite (pglite.dev) as the dev database? 

Pros:
1) Less magic/astonishment ("wait, what database am I using rn?")
2) Simpler schema definition API (currently stuck with the subset of stuff that both PG and SQLite support)

Building ponder.sh. EVM, TypeScript, open-source enjoyer.

3) No change to the local dev experience (one Node.js process, don't have to run a PG server)

Cons:
1) PGlite is immature
2) Locking into PG could be a bad idea - is it better to stick with the narrow subset approach, and add support for even more databases? (doubt)

Hi! My name is Daniel. I am building bountycaster.xyz

Ex Phantom, 0x Project engineer.

From Italy originally, currently live in NYC

Yes. More precisely - what database is used the very first time someone runs Ponder on their local machine.

It needs to be in-process - if you had to start a PG server to get Ponder working, I estimate ~50% of current users would have churned at that step 🤷‍♂️

When you say as a "dev database" are you referring to local development?

Making investing together better https://onit.chat | Variant Founder Fellow | Cambridge Alum | Formerly Ireland's #3 powerlifter (74kg)

Yea agree, the traditional wisdom of avoiding differences between dev and prod is another "pro".

I would vote for pglite because it is just strange developing an environment that is fundamentally different than what is in production 

We had some issues with this but tbf we are hacking around ponder a bit to achieve what we want so we are more exposed to these issue probably

likes building things • https://alexgrover.me/links

FWIW we ran into a discrepancy between local (sqlite) and production almost immediately when we started using ponder, so ended up setting up postgres for local dev. Assuming pglite is reasonably stable it would be preferable IMO!

/max • /bracket • @quickcast • /coys • maxp.co

The big one is for Ponder to simply use Drizzle (drizzle-orm/pg-core) instead of the custom ponder.schema.ts table definitions and Store API.

Big maintenance burden win for us, and there's a decent chance that new users already know Drizzle (vs a 0% chance they know our custom API 😉 )

Curious for your take!

what features under #2 are unlocked if the switch was made?

Good point. I'm actually not too worried about it though, seems like PGlite is very focused on browser support (WASM-first).

If anything a "con" for PGlite is that it's *too* focused on running in the browser, and they might deprioritize Node.js/Bun (where Ponder is more certain to run).

Sqllite I like because the path to ponder running in the background of a local first app is more clear

I’d vote no. If it ain’t broke don’t fix it. Lots of support for SQLite. Not much for PGlite ( my first time hearing about it is this post )

Sounds like an experiment? Maybe run a testing branch for interested devs

Late to the discussion, but +1 for having exactly the same setup. So needs a docker with postgresql etc for local dev. subsquid does it quite well IMO