it's maybe about 50mb heavier on the RAM but it's worth it

you forgot stage 6: Go back to a real OS you can customise yourself rather than consuming regurgitated crap someone else chewed up

dev.to/vikkio88/oma...

omarchy is just a gimmick, is a pre-configured tiling manager with some preinstalled crap on it.

You better off make it yourself from scratch instead of being forced to use all of that rubbish it installs.
I used it for 4 months and just switched back to what I was using before.

I ve been using in enterprise since 2018 and still don't know if I am using it right 🤣

yes please

those are not mutually exclusive, it's in the gray area, only everyone has a different spectrum of gray. I think if the project is yours you can read suggestion and ignore them as much as you want, it's your project after all and you know what's best for it.

the leaking data structure to the fe also is not really a problem, especially if the db schema is there for you to see in an open source project

that's a better answer, I like to follow good practices but dogma are never good and often a sign of premature optimisation. YAGNI

I see, but if you expose partial mapped dto from the data layer, you will see fine. to me it looks like a bigger problem to maintain two types across fe and db layer

I think that's a bit of a dogmatic sentence, it depends on the data. why would it be a security issue leaking the data structure?

I see, but I think sometimes, if the model is quite small and with clear boundaries there's no reason to map 1:1 props, might as well just use the type inferred from the query/repository layer, especially in a full stack app.

rather than having duplicate names everywhere.

and move the walks methods in there.

about the leaking db, it's a full stack app, you are leaking data already by fetching it and showing it to the users arguably 😂

that's why I haven't fully embraced AI into my editor, I want to understand what I do then delegate boring bits to LLM. Like what I used to do with stack overflow.

I usually make a repository, that exposes functions to manipulate that particular resource, like I would do data/walks.ts

it looks nice, only I see loads of duplication of types, I usually infer the types from the drizzle schema then use it on the FE app, so I can keep it consistent in case I change the schema.

also having all of the queries in one single file makes it harder to maintain.