MiNiMAL Posts

MiNiMAL

•Created by MiNiMAL on 11/2/2023 in #❓-help

Function to release context post-batching?

In our application every customer has their own schema, therefore we need a db client per request which I was hoping to leverage with TRPC context. Ideally, we'd be able to: - Connect to the database in context ✅ - Set the search_path to match the client's JWT in context ✅ - Run our procedures using that database connection from context ✅ - Release the database connection after all batching is finished ❌ Is this not currently possible with TRPC? Right now I think I can only connect to the database using a per procedure middleware.

const db = // connect to db

await db.execute('SET search_path=...')

try {
   return await next({ ctx: { ...opts.ctx, db } })
} catch () {
   // release connection
} finally {
   // release connection
}

const db = // connect to db

await db.execute('SET search_path=...')

try {
   return await next({ ctx: { ...opts.ctx, db } })
} catch () {
   // release connection
} finally {
   // release connection
}

This works, but it requires establishing a connection/search_path for every procedure in a batch, instead of just once in context and releasing after the batch. Is there a way to accomplish this?

13 replies

TtRPC

•Created by MiNiMAL on 10/15/2023 in #❓-help

Dynamic input not leading to changes on the front end

I'm building a dynamic procedure to access my database via Drizzle ORM. Ideally a developer would be able to send input parameters to the procedure and have it's output adjusted by that input. For example, in the below, I have a table called demotable which has an updatedAt column. If a developer passes in columns: { updatedAt: false } to the procedure, then ideally the column would not get included in the result. If I hard code the columns parameter in the backend itself, then it works great. Is there a way to accomplish this for the full round trip?

// Inferring type for the findMany argument
type FindManyArg = Parameters<DB['query']['demotable']['findMany']>[0]

// ...

//Setting up the procedure
discordTest: privateProcedure
  .input(input => input as FindManyArg)
  .query(async ({ input }) => {
    const result = await db.query.demotable.findMany(input)
    return result
}),

// On the front end,
const { data: demo } = trpc.discordTest.useQuery({
  columns: { updatedAt: false }, // <== this is properly typed and intellisense is working
})
console.log(demo?.[0].updatedAt) // <=== does not error

// ============================

// Hardcoding columns does properly reflect on the front end
const result = await db.query.demotable.findMany({
  ...input,
  columns: { updatedAt: false },
})
return result

// On the front end,
const { data: demo } = trpc.discordTest.useQuery({})
console.log(demo?.[0].updatedAt) // <=== errors as expected

// Inferring type for the findMany argument
type FindManyArg = Parameters<DB['query']['demotable']['findMany']>[0]

// ...

//Setting up the procedure
discordTest: privateProcedure
  .input(input => input as FindManyArg)
  .query(async ({ input }) => {
    const result = await db.query.demotable.findMany(input)
    return result
}),

// On the front end,
const { data: demo } = trpc.discordTest.useQuery({
  columns: { updatedAt: false }, // <== this is properly typed and intellisense is working
})
console.log(demo?.[0].updatedAt) // <=== does not error

// ============================

// Hardcoding columns does properly reflect on the front end
const result = await db.query.demotable.findMany({
  ...input,
  columns: { updatedAt: false },
})
return result

// On the front end,
const { data: demo } = trpc.discordTest.useQuery({})
console.log(demo?.[0].updatedAt) // <=== errors as expected

4 replies

TtRPC

•Created by MiNiMAL on 10/11/2022 in #❓-help

Best practices for implementing an offline application

Hey there! I'm building a full stack react-native app with Expo and a tRPC backend. I'd like for this app to be functional offline. Currently I'm accomplishing this via a simple optimistic update like:

const createUser = trpc.userCreate.useMutation({
        onMutate: async () => {
            // Cancel users fetch
            await utils.users.cancel()
            // Get current users list
            const previousData = utils.users.getData() || []
            // Append new user
            utils.users.setData([...previousData, { id: uuid.v1(), name } as User])
            return { previousData }
        },
    })

const createUser = trpc.userCreate.useMutation({
        onMutate: async () => {
            // Cancel users fetch
            await utils.users.cancel()
            // Get current users list
            const previousData = utils.users.getData() || []
            // Append new user
            utils.users.setData([...previousData, { id: uuid.v1(), name } as User])
            return { previousData }
        },
    })

This is working pretty well (it's late and I plan on extensively testing this tomorrow). I'd like to confirm what best practices, if any, have been established by the community before continuing.

2 replies