2023-06-29 - kampus coworking daily notes

[usirin]

Things to do, notes from stream, etc etc

[usirin]

add planetscale

[usirin]

i knew branching idea was cool, but i, again, came to the conclusion without having the enough information.

it allows us to branch out from our "Prod" db in every pr preview :corefuWOA2:

[usirin]

prisma & planetscale best practices requires us to not use prisma migrations when deploying to pscale.
that's fine, we can still use migrations in dev, then our ci/cd pipeline would run the prisma db push.

[usirin]

monorepo stuff

[usirin]

went through the pull requests, merged #466. LETS GOOOOOOO @akselsaatci congratz on your first PR

[usirin]

Another "First Pull Request Merged" achievement from @alicangunduz with #470. LETS GOOOOOOOOOOO

[akselsaatci]

LETSS GOOOOOOOOO 🎉🎉🎉

[usirin]

search????

[puskuruk]

Since we use Prisma and MySQL(PlanetScale) we can check the discussion below. Also Prisma explains how we can use it in their official docs

prisma/prisma#11249

[Ketcap]

+1 with @puskuruk , prisma has good enough full text search for now. Then when it get bigger we can go similiar to Algolia or even Algolia :D

[EmirBoyaci]

I don't know exactly what we're mentioning by "search" but I know meilisearch which is a flexible and powerful user-focused search engine.

Since we use Prisma and MySQL(PlanetScale) we can check the discussion below. Also Prisma explains how we can use it in their official docs

prisma/prisma#11249

^ I agree with @puskuruk we can start by using Prisma's text search support. In the future, if we came up some problems like latency and so on, we may move to the another possible solution tho.

[usirin]

We have pano with db, yes. But sozluk's content is being generated from mdx files and being fed to the gql via contentlayer, which means we don't have a db there. This requires us to have a higher level search abstraction which defers to the lower level search functionalities. (fulltextsearch via db, and in-memory search solution for sozluk for now)

I mostly agree about not using a third party but, the above situation is making me think that maybe we should start doing it from the beginning.

[Ketcap]

@usirin what about turning content into vectors and do json vector search ? convertion on vector can be done on build time