Today, we are issuing the 3.11.1 patch release.

MongoDB (Preview)

Breaking: Filters no longer return undefined fields by default

In 3.11.1, we've changed what data is returned when filtering MongoDB documents on undefined fields. The new rule is that undefined fields are excluded by default unless explicitly filtered for. This allows you to query for undefined and null values separately.

Let's take a look at a concrete example. Given the following Prisma schema:

model Address {
    id     Int    @id @map("_id")
    city   String
    street String? // Note that street is optional
}

For Mongo, optional fields can either be null or undefined (absent). The following documents are all valid for the schema above:

{ "_id": 1, "city": "San Fransisco", "street": "Market st." }
{ "_id": 2, "city": "Seattle", "street": null }
{ "_id": 3, "city": "Chicago" }

Prior to 3.11.1, if you queried for where: { street: null }, you'd get _id: 2 and _id: 3. In 3.11.1, you'll only get _id: 2. The ability to also query for the missing fields has also been added. For details, refer to the new isSet below to learn more.

There are a few exceptions to this new default:

• A having filter on an aggregated field will return undefined fields. This is because aggregation on undefined fields yields null, not undefined, thus matching the filter.
• Filters on undefined to-many relations (e.g., the backing array of a many-to-many is undefined) will currently include those relations in the result set.

New isSet filter operation

To compensate for missing fields on documents no longer being returned by the filters above, we’ve added a new isSet: bool filter. This filter can be used to include fields that are undefined on documents.

Using the example above, to include the undefined fields, you can use an OR:

await prisma.address.findMany({
  where: {
    OR: [
      { street: { isSet: false } },
      { street: null }
    ]
  }
})

The isSet operation has been added to all scalar and embedded fields that are optional.

New unset operation

In 3.11.1, you can also remove a field with the unset operation.

Using the example above, let's write a query to remove the street field:

await prisma.address.update({
  where: {
    id: 10,
  },
  data: {
    street: {
      unset: true,
    },
  },
})

This effectively sets the street field to undefined in the database.

New updateMany operation

We now support updating embedded documents that match specific criteria.

For example, given the following schema:

model Product {
  id          Int  @id @map("_id")
  name        String  @unique
  photos      Photo[]
}

type Photo {
  height Int    @default(200)
  width  Int    @default(100)
  url    String
}

Let's update the photo with a url of 1.jpg to 2.png:

const product = prisma.product.update({
  where: {
    id: 10,
  },
  data: {
    photos: {
      updateMany: {
        where: {
          url: '1.jpg',
        },
        data: {
          url: '2.png',
        },
      },
    },
  },
})

New deleteMany operation

Similar to updateMany, you can also remove embeds that match specific criteria.

Using the Prisma Schema above, let's delete all photos with a height of 100:

const product = prisma.product.update({
  where: {
    id: 10,
  },
  data: {
    photos: {
      deleteMany: {
        where: {
          height: 100,
        },
      },
    },
  },
})

Today, we are excited to share the 3.11.0 stable release 🎉

🌟 Help us spread the word about Prisma by starring the repo or tweeting about the release. 🌟

Major improvements and new features

Experimental support for Embedded Document Filters

In the previous release, we added embedded document support for creates, updates, and deletes. In version 3.11.0, we’re adding the ability to filter embedded documents.

Given the following schema:

datasource db {
  provider = "mongodb"
  url      = env("DATABASE_URL")
}

generator client {
  provider        = "prisma-client-js"
  previewFeatures = ["mongoDb"]
}

model Product {
  id     String  @id @default(auto()) @map("_id") @db.ObjectId
  photos Photo[]
}

model Order {
  id              String   @id @default(auto()) @map("_id") @db.ObjectId
  shippingAddress Address
  billingAddress  Address?
}

type Photo {
  height Int
  width  Int
  url    String
}

type Address {
  street String
  city   String
  zip    String
}

You can now filter within an embedded document:

// find all orders with the same shipping address
const orders = await prisma.order.findMany({
  where: {
    shippingAddress: {
      equals: {
        street: "555 Candy Cane Lane",
        city: "Wonderland",
        zip: "52337",
      },  
    },
  },
})

You can also filter on a "contains many" relationship:

// find all products that don't have photos
const product = prisma.product.findMany({
  where: {
    photos: {
      isEmpty: true
    }
  },
})

This scratches the surface of what's possible. For a complete list of available operations, have a look at our documentation. Please share your feedback in this issue.

Ordering by embedded documents is in Preview

In addition to filtering, Prisma version 3.11.0 now supports sorting by an embedded document.

Using the example schema above, you can sort orders by their zip code:

// sort orders by zip code in ascending order
const orders = await prisma.order.findMany({
  orderBy: {
    shippingAddress: {
      zip: "asc",
    },
  },
})

Learn more about this feature in our documentation and don’t hesitate to reach out in this issue.

MongoDB query logging support

In this release, we’ve added the ability to log MongoDB queries. You can enable query logging in the PrismaClient constructor:

const prisma = new PrismaClient({
  log: [
    {
      emit: 'event',
      level: 'query',
    },
  ]
})

prisma.$on('query', (e) => console.log(e.query))

After enabling query logging, you'll start to see logs that resemble this in your console:

db.User.deleteMany({ _id: ( $in: [ “62261e0b18139c6099ba7097”, ], }, }) 
db.User.deleteMany({ _id: ( $in: [ “6226277a96069500743edcf9”, ], }, })

The logs output by Prisma have the same format as the mongosh console, so you can pipe the queries from your logs directly into your shell.

MongoDB introspection update

We've updated the type inference behavior for MongoDB on introspection.

Prisma samples a field's data to select an appropriate type on introspection. In the past, Prisma picked the type used most often for fields with data with multiple types. However, this could cause problems when retrieving mixed data during runtime and throw exceptions, such as Prisma Studio or in Prisma Client queries.

From 3.11.0, Prisma defaults to the Json type to all fields with mixed data types instead. Additionally, Prisma will still show a warning on the console and add a comment to the introspected Prisma schema so it is clear where such cases occur and that you can do something to fix them.

Prisma Client logger revamp

In 3.11.0, we’ve rewritten our internal logger to reduce lock contention and enable future features like tracing. This is the first of many upcoming changes to improve the Prisma Client’s throughput, so if you were running into an upper limit on query performance, it’s time to update Prisma Client and give it a try!

If you're running into query performance issues, please open an issue or connect with us on Slack.

CockroachDB now supports migrations (Preview)

We're excited to announce Preview support for migrations for CockroachDB. You can now evolve your Prisma schema and propagate the changes to your database using Prisma Migrate.

Give the feature a try and let us know what you think in this issue.

Detecting state of a diff with migrate diff using exit code

Prisma version 3.11.0 includes a new --exit-code flag to the migrate diff command to detect the state of a diff in several ways.

You can use the flag as follows:

npx prisma migrate diff --preview-feature \
--exit-code \
--from-[...] \
--to-[...]

Here's a list of the default and changed behavior of the error codes:

## Default behavior of exit codes
0: Returned when the diff is empty or non-empty
1: Returned on error

## Changed behavior when --exit-code is used
0: Returned when the diff is empty
1: Returned on error
2: Returned when the diff is non-empty

Read about it in the reference documentation.

Fixes and improvements

Prisma

• Command to export SQL schema
• Waiting "too long" to input migration name leads to error message
• db push with empty schema.prisma: Error: TypeError: Cannot read properties of undefined (reading 'url')
• migrate diff: --{from,to}-schema-datamodel should not error on missing env vars in datasource blocks
• migrate diff: Provide a reliable way to detect empty diffs / migrations
• db execute cannot resolve SQLite file path from schema
• [MDB] Implement version and describe RPC calls for the error reporting backend
• [MDB] Create fields referenced in index definitions as Json to enable display of indexes in datamodel
• MongoDB many to many: PANIC: called Option::unwrap() on a None

Prisma Client

• PANIC in query-engine/connectors/mongodb-query-connector/src/value.rs:185:24not yet implemented: (Json, Json("[]"))
• Prisma Mongo Panic called Option::unwrap() on a None value
• Better error message for MongoDB replica sets
• MongoDB: Using tlsCAFile fails
• no such file or directory, open '/path/schema.prisma' since 3.1.1
• PANIC: called Option::unwrap() on a None value in query-engine/connectors/mongodb-query-connector/src/root_queries/write.rs:301:74
• contains string filter not working with mongoDB
• Referential integrity is not preserved when updating foreign key (in "prisma" mode)
• PANIC: called Option::unwrap() on a None value in query-engine\connectors\mongodb-query-connector\src\root_queries\read.rs:112:74
• HasReject not evaluating correctly for per operation handlers
• Support MongoDB "query" logging
• CONTRIBUTING: document Prisma Client workflow of using local link
• MongoDB findRaw filter problem with ObjectId field.
• MongoDB: Better error message for when document does not match the defined schema
• Integrate orderBy for composite types
• Attempting to do a query in the Prisma Client on a model with a foreign key that has @map panics on 3.10.0
• Use embedded documents attributes in where clause
• 'then' in PrimsaPromise returns false

Prisma Migrate

• Detect schema drift helper / migrate dev --exit-code
• Migrate lock timeout on PlanetScale

Language tools (e.g. VS Code)

• MongoDB M:N relations get referential actions auto completion suggestions, then does not validate
• Auto Completion: map does not exist for M:N in MongoDB
• Autocomplete shows invalid suggestions for composite types
• Suggest auto() in VSCode autocomplete
• Clicking on a MongoDB composite type does not jump to definition

Prisma Engines

• Implement order by composites
• [Composites] Implement equals read operation
• [Composites] Implement is read operation
• [Composites] Implement isNot read operation
• [Composites] Implement isEmpty read operation
• [Composites] Implement every read operation
• [Composites] Implement some read operation
• [Composites] Implement none read operation

Credits

Huge thanks to @hayes, @maddhruv, @jasimon for helping!

📺 Join us for another "What's new in Prisma" livestream

Learn about the latest release and other news from the Prisma community by joining us for another "What's new in Prisma" livestream.

The stream takes place on YouTube on Thursday, March 17 at 5 pm Berlin | 8 am San Francisco.