Question on deploying to runway

[aaa3334]

Is it preferable when deploying to runway (https://docs.medusajs.com/resources/deployment/medusa-application/railway) to:

1. Run medusa build locally, remove .medusa from the .gitignore then follow the info in the guide (note those first two are not mentioned in the railway guide) or
2. Use the start command: npm run build && cd .medusa/server && npm run prebuild && npm run start
   Without the .medusa folder (which is in the default .gitignore) we can't run the given start command which is cd .medusa/server && npm run prebuild && npm run start

Either way the docs have a gap but am wondering which way is preferable?

[aaa3334]

For anyone wondering,

--- Root directory ---

1. if you are using the nextjs template, you should specify the base path in the 'Root directory' path in settings.

--- Build ---
2) Change your build command to: "build": "npx medusa build" (I tried the 'admin only' one but it actually just built the front end not backend which was odd) (can write 'npm run build' in the part on railway)

--- Deploy ---
(Note I connected up supabase - maybe not ideal but I skipped the prebuild as I had already set that up - but I did have to run npm install for it to work)
3) cd .medusa/server && npm install && npm run start
(in package.json its: "start": "npx medusa start")