c243ded6-9e79-4752-857d-5b6f31fd8cc2

You can define a packageManager in your package.json, but if that doesn't work for you then you can also override the install and build command on Railpack via RAILPACK_INSTALL_CMD and RAILPACK_BUILD_CMD

I figured it out. Created a nixpacks.toml file in my backend:

[phases.setup]
nixPkgs = ["nodejs", "pnpm"]

[phases.install]
cmds = ["pnpm install"]

[phases.build]
cmds = ["pnpm run build"]

[start]
cmd = "pnpm run start:prod"

This did the job.

If this worked for the apps/backend, it should work for apps/frontend too

I suppose that works but just remember that wouldn't be cross compatible with Railpack if you ever switch in the future

Is one better than the other? Right now I just set out with the default configurations (AI helped me with the nixpack stuff)

Nixpacks is Railway's first generation builder and Railpack is the replacement second generation builder, from my understanding Nixpacks isn't being worked on anymore and Railway is actually already in the process of defaulting over to Railpack (only for some new services)

Setting pnpm as your packageManager in your package.json rather than using nixpacks.toml will work for both Railpack and Nixpacks so if you ever need to switch in the future then you don't need to translate Nixpacks configurations

Thank you. I'll check this out.

An unrelated question to the topic but can I not install the vector extension on my postgres instance?

I don't think so, but you can install a different postgres database that has the vector extension preinstalled

(made by a Railway team member)

I've got one more small thing. As stated that it's a pnpm monorepo structure. I've set the root directory for both as well:

apps/frontend and apps/backend

And I've set the watch paths for both like this:

apps/backend/src/** and apps/frontend/src/**

What I thought would happen is that deployment for each app would happen based on whether I have made changes in their respective paths. But right now I edited the nixpacks.toml in the backend folder and it triggered deployments for both. What am I missing here?

That's weird <:Thinking:1360710341239242762> I didn't think it'd do that, but I also don't use monorepo so I guess I never had a chance to find out that I'd be wrong

regardless, you can set up watch paths for each service to tell Railway which paths should change before triggering a new deployment, this will be in your service settings

Oh lol nvm it worked

I am stupid... it actually skipped 🤣

"No changed files matched patterns:"

awesome!!

all good to close this thread?

Yes. Do I close this or will an admin do that

I'll do that no worries

thank youuu

you will? 😆

trust

!s

lol

i have to let it marinate