Migrating from Postgres to SQLite with Kent C. Dodds

[@aarondfrancis]

You can also listen to this episode as a podcast on your favorite podcast player: databaseschool.transistor.fm/episodes/migrating-from-postgres-to-sqlite-with-kent-c-dodds

[@zuma206]

this series really is liquid gold. just took kent's litefs/global distribution course, now he's on the podcast. love it

[@wesleycoder]

Aaron and Kent: such a chill duo, love to see it. I liked the name "database schools" sounds interesting, and I bet it will attract attention. I'll be spending the rest of my day now thinking about the curse of knowledge...

[@ajinkyax]

I was looking for Kent C. Dodds talk about Remix and SQLite at production 1 year old talk and I found this. :)

[@MrDadidou]

Aaron, thank you so much for your enthusiasm, your work and your positivity!

[@pookiepats]

Thank you for existing, breathing & eating!

[@moodyhamoudi]

Positive energy overload with these two

[@rembautimes8808]

I did try epic web accessing it from Malaysia and I was noticing that the site loads really really fast. Great talk 😂

[@sean_reyes]

does SQlite have good Json Column support?

[@theplaintech]

Yes, however, libsql is a better choice, especially for JSON.

[@relaxwithai]

Kinda, from my understanding, you can use the TEXT datatype and perform json operations on it

[@benlevy1896]

It has both json that is stored as text and jsonb that gets stored as binary.

[@HideBuz]

​@@theplaintech Why? How did they improve json in libsql?

[@theplaintech]

@@HideBuz For full effect and edification, do your own research.

[@popetgirl]

A podcast with pocketbase team. They are using sqlite.

[@jit-r5b]

Team? Wasn't it one amazing dev from Bulgaria that built it?

[@MrDpof]

Yes!!!! It would awesome.

[@pookiepats]

it is one dev lol the only thing he outsourced was the websites css 😂

[@7ala9at]

can i use sqlite in production now?

[@PhilippeLoctaux]

i would love to be able to subscribe to this show in my podcast client, when is the rss feed coming? :)

[@aarondfrancis]

It's here! databaseschool.transistor.fm

[@PhilippeLoctaux]

thanks! you might want to put it in the description so more people can find it!

[@DanielTolentino42]

​@@PhilippeLoctaux +1

[@gregorywpower]

One thing I think would be even more bananas would be replacing that caching layer with DuckDB.

[@nexovec]

This is really silly, but I need LISTEN/NOTIFY.

[@FaraazAhmad]

So Kent is your Aaron Francis

[@versaleyoutubevanced8647]

good to see other ecosystems have their own kent

[@ordinarygg]

Geodata and a lot of missing features unfortunately SQLite is not for big production. SQLIte is great for single server projects and hobby one.

[@antidegenerates7449]

Not every big project needs geodata 🤡 and for geodata theres dedicated solutions

[@ordinarygg]

@@antidegenerates7449 I think 🤡 is using half-compatible databases for any complex data, for example JSON. Do you really think single file will be faster in random access data then inode splited where OS proper cache is. Imaginary ponies that don’t have 10bil table in their lives, calling something “big projects” lol

[@Jason-xw2md]

​@@ordinaryggthe vast majority of businesses are not going to have 10 billion rows. real 🤡 is acting like one tool is always the "correct" solution

[@pookiepats]

@@antidegenerates7449exactly, either way you're going over the wire for geodata so his comment is pure cult driven criticism, as if "big projects" only use a single data store.

[@nomadshiba]

skill issue. sqlite is easier to scale and distribute. geodata is just an abstraction, not a can't do without feature. it's just numbers, latitude and longitude are just floating point values. you can store them in sqlite just fine and query them efficiently if you use the right indexing method, like r-trees or simple sorted indexes with range queries. games have been doing spatial partitioning forever without "geospatial databases." they use grids, quadtrees, octrees, spatial hashing, and sector based chunking. it's just about organizing data in a way that makes lookups fast. same goes for "vector databases". if you have to scale, you can move to sqlite and distribute your databases to hundreds, thousands, millions of databases. its just a file and you can have more than one file at more than one location. file per user, file per group, etc. you can have protocols on how database of each user interacts with eachother when users interact. you can handle millions, billions of total read and write per, second. scale to infinity with no bottleneck.