Today, we are excited to share the 5.15.0 stable release 🎉

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

Highlights

Multi-File Prisma Schema support

Prisma ORM 5.15.0 features support for multi-file Prisma Schema in Preview.

This closes a long standing issue and does so in a clean and easy to migrate way.

To get started:

1. Enable the prismaSchemaFolder Preview feature by including it in the previewFeatures field of your generator.
   datasource db {
   provider = "postgresql"
   url = env("DATABASE_URL")
   }

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

2. Create a schema subdirectory under your prisma directory.
3. Move your schema.prisma into this directory.

You are now set up with a multi-file Prisma Schema! Add as many or as few .prisma files to the new prisma/schema directory.

When running commands where a Prisma Schema file is expected to be provided, you can now define a Prisma Schema directory. This includes Prisma CLI commands that use the --schema option as well as defining schema via package.json

Our tooling has also been updated to handle multiple Prisma Schema files. This includes our Visual Studio Code extension and tools like database introspection, which will deposit new models in a introspected.prisma file. Existing models will be updated in the file they are found.

To learn more, please refer to our official documentation and announcement blog post. If you try out prismaSchemaFolder, please let us know!

Interesting Bug Fixes

Fix for PostgreSQL prepared statement caching for raw queries

This release fixes a nasty bug with the caching of prepared statements in raw Prisma Client queries that affected PostgreSQL when you ran the same SQL statement with differently typed paramters. This should not fail any more.

Fix for SQL Server introspection of (deprecated) CREATE DEFAULT

Our Introspection logic crashed on encountering certain multi-line CREATE DEFAULT, a deprecated way to define defaults in SQL Server. As many SQL Server users are working with established databases, this happened frequently enough that we now explicitly ignore these defaults instead of crashing.

Fix for Cloudflare D1’s lower parameter limit

Cloudflare’s D1 has a lower parameter limit than local SQLite, which caused bigger queries to fail. We adapted that limit to the D1 default for @ prisma/adapter-d1, which will avoid such failures.

Fix for Cloudflare D1’s different PRAGMA support

Our generated migration SQL for SQLite did not always work for Cloudflare D1, because of differences in the supported pragmas. We adapted the SQL to work in both local SQLite and Cloudflare D1.

Fixes and improvements

Prisma Migrate

• Crash on multiline defaults introspection on MSSQL
• Error: [libs\sql-schema-describer\src\mssql.rs:315:30] called Result::unwrap() on an Err value: "Couldn't parse default value: create default [dbo].[member_notification_cancel_flags] as 0\r\n"
• Error: [libs\sql-schema-describer\src\mssql.rs:315:30] called Result::unwrap() on an Err value: "Couldn't parse default value: create default d_password as 'D,73' "
• Crash introspecting MSSQL database with DEFAULTs
• doing introspection on a SQL Server 2018 DB - for Dynamic GP get the following error.
• Error: [libs\sql-schema-describer\src\mssql.rs:317:30] called Result::unwrap() on an Err value: "Couldn't parse default value: \r\ncreate default D_BIT_OFF\r\nas 0\r\n"
• Error: called Result::unwrap() on an Err value: "Couldn't parse default value in SQL Server
• db pull errors on SQL Server with Error: [libs\sql-schema-describer\src\mssql.rs:336:30] called Result::unwrap()on anErr value: "Couldn't parse default value: [...]
• Error: [libs\sql-schema-describer\src\mssql.rs:336:30] called Result::unwrap() on an Err value: "Couldn't parse default value: \r\ncreate default [va_nulla] as 0\r\n"
• Error when pulling from database
• Foreign key relation results in erroneous second migration
• db pull can't parse script setting default value
• Bug: Migrations not compatible with D1
• SQL Server Introspection crashes on multi-line (deprecated) defaults

Prisma Client

• Raw query failed. Code: 22P03. Message: db error: ERROR: incorrect binary data format in bind parameter 1
• Float number on raw query: incorrect binary data format in bind parameter 1
• Can't use Prisma client in Next.js middleware, even when deploying to Node.js
• Prepared statement caching is data dependent on numeric input parameters (incorrect binary data format in bind parameter x)
• Turso Driver Adapter: Including _count leads to error
• Next.js app build fails when using Prisma with DB driver in Server Action
• Bug: [D1] Error in performIO: Error: D1_ERROR: too many SQL variables
• Remove warn(prisma-client) This is the 10th instance of Prisma Client being started. warning in Edge (and potentially) other envs)
• $executeRawUnsafe: incorrect binary data format in bind parameter 6
• Bug: Error or bug using Prisma with DriverAdapter with PostgreSQL database Neon
• Inconsistent column data: Unexpected conversion failure from Number to BigInt error when using @prisma/adapter-pg
• Incompatibility with NextJS app dir, CloudFlare Pages and D1
• Breaking change? Int switched to being Int32 for MongoDB

Language tools (e.g. VS Code)

• VS Code extension is showing an advertisement
• Generate codelens fails on Windows
• We incorrectly read commented out preview features if they are before the real preview features

Credits

Huge thanks to @ pranayat, @ yubrot, and @ skyzh for helping!

Today, we are excited to share the 5.14.0 stable release 🎉

🌟 Help us spread the word about Prisma by starring the repo ☝️ or posting on X about the release. 🌟

Highlights

Share your feedback about Prisma ORM

We want to know how you like working with Prisma ORM in your projects! Please take our 2min survey and let us know what you like or where we can improve 🙏

createManyAndReturn()

We’re happy to announce the availability of a new, top-level Prisma Client query: createManyAndReturn(). It works similarly to createMany() but uses a RETURNING clause in the SQL query to retrieve the records that were just created.

Here’s an example of creating multiple posts and then immediately returning those posts.

Additionally,createManyAndReturn() supports the same options as findMany(), such as the ability to return only specific fields.

Full documentation for this feature can be found in the Prisma Client API Reference.

Note: Because createManyAndReturn() uses the RETURNING clause, it is only supported by PostgreSQL, CockroachDB, and SQLite databases. At this time, relationLoadStrategy: join is not supported in createManyAndReturn() queries.

MongoDB performance improvements

Previously, Prisma ORM suffered from performance issues when using the in operator or when including related models in queries against a MongoDB database. These queries were translated by the Prisma query engine in such a way that indexes were skipped and collection scans were used, leading to slower queries especially on large datasets.

With 5.14.0, Prisma ORM now rewrites queries to use a combination of $or and $eq operators, leading to dramatic performance increases for queries that include in operators or relation loading.

Fixes and improvements

Prisma Client

• createMany() should return the created records
• Generating Prisma client without any model in its schema
• MongoDB: Performance issue with nested take on many-to-one relationship
• Slow queries on MongoDB using include for relations
• [MongoDB] Performance issue in findMany() query execution with in
• MongoDB nested/include query slow
• MongoDB Connector generates queries which do not take advantage of indices.
• Mongodb Nested Queries Not Using Indexes
• MongoDB slow delete with onDelete: SetNull
• Slow query with many-to-one relationship on MongoDB
• prisma init --with-model
• Fixed version of @ opentelemetry/* dependencies
• Usage of deprecated punycode module
• Bug: D1 One-to-Many Relation INSERTs fail with The required connected records were not found. when using indices

Prisma Migrate

• Empty dbgenerated() still breaking for Unsupported() types
• Validation error when shadowDatabaseUrl is identical to url (or directUrl)
• MongoDB Query with 'in' condition will cause COLLSCAN, without leveraging indexes
• "Not Authorised" when directly applying Prisma generated migrations to Cloudflare D1 with PRAGMA foreign_key_check;

Language tools (e.g. VS Code)

• make superior: model generate
• Missing code autocomplete for referential actions with mongodb

Company news

Prisma Changelog

Curious about all things Prisma? Be sure to check out the Prisma Changelog for updates across Prisma's products, including ORM, Accelerate, and Pulse!

New product announcement: Prisma Optimize

With this release, we are excited to introduce a new Prisma product. We’re calling it “Optimize” because that’s what it does! Let your favorite ORM also help you debug the performance of your application.

Check out our announcement blog post for more details, including a demo video.

Credits

Huge thanks to @ pranayat, @ yubrot, @ skyzh, @ anuraaga, @ gutyerrez, @ avallete, @ ceddy4395, @ Kayoshi-dev for helping!

Today, we are excited to share the 5.13.0 stable release 🎉

🌟 Help us spread the word about Prisma by starring the repo or posting on X about the release.

Highlights

omit fields from Prisma Client queries (Preview)

We’re excited to announce Preview support for the omit option within the Prisma Client query options. The highly-requested omit feature now allows you to exclude fields that you don’t want to retrieve from the database on a per-query basis.

By default, when a query returns records, the result includes all scalar fields of the models defined in the Prisma schema. select can be used to return specific fields, while omit can now be used to exclude specific fields. omit lives at the same API level and works on all of the same Prisma Client model queries as select. Note, however, that omit and select are mutually exclusive. In other words, you can’t use both in the same query.

To get started using omit, enable the omitApi Preview feature in your Prisma schema:

// schema.prisma
generator client {
  provider        = "prisma-client-js"
  previewFeatures = ["omitApi"]
}

Be sure to re-generate Prisma Client afterwards:

npx prisma generate

Here is an example of using omit:

// Includes all fields except password
await prisma.user.findMany({
  omit: {
   password: true
  },
})

Here is an example of using omit with include:

// Includes all user fields except user's password and title of user's posts
await prisma.user.findMany({
  omit: {
   password: true
  },
  include: {
    posts: {
      omit: {
        title: true
      },
    },
  },
})

Many users have requested a global implementation of omit. This request will be accommodated in the future. In the meantime, you can follow the issue here.

📣 Share your feedback: omitApi Preview feature

📚 Documentation: omit - Prisma Client API Reference

Fixes and improvements

Prisma Migrate

• Nicer Datamodel Type validation

Prisma Client

• MySQL upsert(): Internal error: Attempted to serialize empty result.
• Mysql upsert() fails with "Attempted to serialize empty result."
• Interactive Transaction API timing out after upgrading from 4.7.0 and above
• MySQL and SQLite upsert(): Internal error: Attempted to serialize empty result.
• MySQL upsert(): Internal error: Attempted to serialize empty result.
• MongoDB upsert(): Internal error: Attempted to serialize empty result.
• MongoDB upsert(): Internal error: Attempted to serialize empty result
• Relation mode + upsert(): Internal error: Attempted to serialize empty result.
• Internal error: Attempted to serialize empty result. on upsert() for update case in different databases (when using relationMode=prisma explicitly or implicitly [MongoDB])
• Postgres upsert(): Internal error: Attempted to serialize empty result when relationMode = "prisma" is used
• ✘ [ERROR] near "��": syntax error at offset 0 when running wrangler d1 migrations apply with Prisma generated migration (on Windows, using Powershell)

Credits

Huge thanks to @ ospfranco, @ pranayat, @ yubrot, @ skyzh, @ anuraaga, @ yehonatanz, @ arthurfiorette, @ elithrar, @ tockn, @ Kuhave, @ obiwac for helping!

Today, we are issuing the 5.12.1 patch release to fix two small problems with our new Cloudflare D1 support.

Fixes in Prisma CLI

Windows-only fix for new D1 specific flags for migrate diff and db pull

The flags --from-local-d1 and --to-local-d1 for migrate diff and --local-d1 to db pull we added in 5.12.0 were not working as expected when running on Windows only. This is now fixed.

📚 Documentation: Deploying a Cloudflare worker with D1 and Prisma ORM

New option for migrate diff: -o or --output

We added a new parameter --output to migrate diff that can be used to provide a filename into which the output of the command will be written. This is particularly useful for Windows users, using PowerShell, as using > to write into a file creates a UTF-16 LE file that can not be read by wrangler d1 migrations apply. Using this new option, this problem can be avoided:

npx prisma migrate diff --script --from-empty --to-schema-datamodel ./prisma/schema.prisma --output ./schema.sql

Related issues:

• ✘ [ERROR] near "��": syntax error at offset 0 when running wrangler d1 migrations apply with Prisma generated migration (on Windows, using PowerShell) #23702
• prisma migrate resolve --applied not working on new project, migration ... could not be found.

Today, we are excited to share the 5.12.0 stable release 🎉

🌟 Help us spread the word about Prisma by starring the repo or posting on X about the release.

Highlights

Cloudflare D1 (Preview)

This release brings Preview support for