v5.21.1

Compare Source

• Fixed a bug where migrations were not using shadow database correctly in some edge cases

v5.21.0

Compare Source

Today, we are excited to share the 5.21.0 release 🎉

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

Highlights

Better support for tracing in MongoDB

The tracing Preview feature now has full support for MongoDB with previously missing functionality now implemented. This is a part of the ongoing effort to stabilize this Preview feature and release it in General Availability.

tracing is a Preview feature that enables built-in support for OpenTelemetry instrumentation inside the Prisma Client and provides deep insights into the performance and timing of your queries. See our documentation for more information.

For an easy to use and zero-configuration tracing instrumentation tool with a dashboard that provides an overview of your queries, statistics, and AI-powered recommendations, try Prisma Optimize.

WebAssembly engine size decrease for edge functions

Due to recent changes, some users experienced a steep increase of the bundle size in Prisma 5.20 when using the driverAdapters Preview feature, going over the 1 MB limit on the free tier of Cloudflare Workers. This has now been fixed.

Fixes and improvements

Prisma Engines

• Avoid regex crate for wasm bundle size (to keep cloudflare free plan)

Credits

Huge thanks to @​austin-tildei, @​LucianBuzzo, @​mcuelenaere, @​pagewang0, @​key-moon, @​pranayat, @​yubrot, @​skyzh for helping!

v5.20.0

Compare Source

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

Highlights

strictUndefinedChecks in Preview

With Prisma ORM 5.20.0, the Preview feature strictUndefinedChecks will disallow any value that is explicitly undefined and will be a runtime error. This change is direct feedback from this GitHub issue and follows our latest proposal on the same issue.

To demonstrate the change, take the following code snippet:

prisma.table.deleteMany({
  where: {
    // If `nullableThing` is nullish, this query will remove all data.
    email: nullableThing?.property,
  }
})

In Prisma ORM 5.19.0 and below, this could result in unintended behavior. In Prisma ORM 5.20.0, if the strictUndefinedChecks Preview feature is enabled, you will get a runtime error instead:

Invalid \`prisma.user.findMany()\` invocation in
/client/tests/functional/strictUndefinedChecks/test.ts:0:0
  XX })
  XX 
  XX test('throws on undefined input field', async () => {
→ XX   const result = prisma.user.deleteMany({
         where: {
           email: undefined
                  ~~~~~~~~~
         }
       })
Invalid value for argument \`where\`: explicitly \`undefined\` values are not allowed."

We have also introduced the Prisma.skip symbol, which will allow you to get the previous behavior if desired.

prisma.table.findMany({
  where: {
    // Use Prisma.skip to skip parts of the query
    email: nullableEmail ?? Prisma.skip
  }
})

From Prisma ORM 5.20.0 onward, we recommend enabling strictUndefinedChecks, along with the TypeScript compiler option exactOptionalPropertyTypes, which will help catch cases of undefined values at compile time. Together, these two changes will help protect your Prisma queries from potentially destructive behavior.

strictUndefinedChecks will be a valid Preview feature for the remainder of Prisma ORM 5. With our next major version, this behavior will become the default and the Preview feature will be “graduated” to Generally Available.

If you have any questions or feedback about strictUndefinedChecks, please ask/comment in our dedicated Preview feature GitHub discussion.

typedSql bug fix

Thank you to everyone who has tried out our typedSql Preview feature and provided feedback! This release has a quick fix for typescript files generated when Prisma Schema enums had hyphens.

Fixes and improvements

Prisma

• Prisma incorrectly parses CRDB's FK constraint error as not available.
• Invalid TypeScript files created by generate when typedSql is enabled and enum contains hyphens.
• @prisma/internals didn't list ts-toolbelt in dependencies.
• using $extends prevents model comments from being passed to TypeScript

Prisma Engines

• Planetscale engine tests: interactive_tx
• Fix broken engine size publishing workflow

Credits

Huge thanks to @​mcuelenaere, @​pagewang0, @​key-moon, @​pranayat, @​yubrot, @​thijmenjk, @​mydea, @​HRM, @​haaawk, @​baileywickham, @​brian-dlee, @​nickcarnival, @​eruditmorina, @​nzakas, and @​gutyerrez for helping!

v5.19.1

Compare Source

Today, we are issuing the 5.19.1 patch release.

What's Changed

We've fixed the following issues:

• https://github.com/prisma/prisma/issues/25103
• https://github.com/prisma/prisma/issues/25137
• https://github.com/prisma/prisma/issues/25104
• https://github.com/prisma/prisma/issues/25101

Full Changelog: prisma/prisma@5.19.0...5.19.x, prisma/prisma-engines@5.19.0...5.19.x

v5.19.0

Compare Source

Today, we are excited to share the 5.19.0 stable release 🎉

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

Highlights

Introducing TypedSQL

TypedSQL is a brand new way to interact with your database from Prisma Client. After enabling the typedSql Preview feature, you’re able to write SQL queries in a new sql subdirectory of your prisma directory. These queries are then checked by Prisma during using the new --sql flag of prisma generate and added to your client for use in your code.

To get started with TypedSQL:

1. Make sure that you have the latest version of prisma and @prisma/client installed:

   npm install -D prisma@latest
   npm install @​prisma/client@latest
   

2. Enable the typedSql Preview feature in your Prisma Schema.

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

3. Create a sql subdirectory of your prisma directory.

   mkdir -p prisma/sql
   

4. You can now add .sql files to the sql directory! Each file can contain one sql query and the name must be a valid JS identifier. For this example, say you had the file getUsersWithPosts.sql with the following contents:

   SELECT u.id, u.name, COUNT(p.id) as "postCount"
   FROM "User" u
   LEFT JOIN "Post" p ON u.id = p."authorId"
   GROUP BY u.id, u.name

5. Import your SQL query into your code with the @prisma/client/sql import:

      import { PrismaClient } from '@​prisma/client'
      import { getUsersWithPosts } from '@​prisma/client/sql'
   
      const prisma = new PrismaClient()
   
      const usersWithPostCounts = await prisma.$queryRawTyped(getUsersWithPosts)
      console.log(usersWithPostCounts)

There’s a lot more to talk about with TypedSQL. We think that the combination of the high-level Prisma Client API and the low-level TypedSQL will make for a great developer experience for all of our users.

To learn more about behind the “why” of TypedSQL be sure to check out our announcement blog post.

For docs, check out our new TypedSQL section.

Bug fixes

Driver adapters and D1

A few issues with our driverAdapters Preview feature and Cloudflare D1 support were resolved via https://github.com/prisma/prisma-engines/pull/4970 and https://github.com/prisma/prisma/pull/24922

• Mathematic operations such as max, min, eq, etc in queries when using Cloudflare D1.
• Resolved issues when comparing BigInt IDs when relationMode="prisma" was enabled and Cloudflare D1 was being used.

Joins

• https://github.com/prisma/prisma/issues/23742 fixes Prisma Client not supporting deeply nested some clauses when the relationJoins Preview feature was enabled.

Join us

Looking to make an impact on Prisma in a big way? We're now hiring engineers for the ORM team!

• Senior Engineer (TypeScript): This person will be primarily working on the TypeScript side and evolving our Prisma client. Rust knowledge (or desire to learn Rust) is a plus.
• Senior Engineer (Rust): This person will be focused on the prisma-engines Rust codebase. TypeScript knowledge (or, again, a desire to learn) is a plus.

Credits

Huge thanks to @​mcuelenaere, @​pagewang0, @​Druue, @​key-moon, @​Jolg42, @​pranayat, @​ospfranco, @​yubrot, @​skyzh for helping!

v5.18.0

Compare Source

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

Highlights

Native support for UUIDv7

Previous to this release, the Prisma Schema function uuid() did not accept any arguments and created a UUIDv4 ID. While sufficient in many cases, UUIDv4 has a few drawbacks, namely that it is not temporally sortable.

UUIDv7 attempts to resolve this issue, making it easy to temporally sort your database rows by ID!

To support this, we’ve updated the uuid() function in Prisma Schema to accept an optional, integer argument. Right now, the only valid values are 4 and 7, with 4 being the default.

model User {
  id   String @​id @​default(uuid()) // defaults to 4
  name String
}

model User {
  id   String @​id @​default(uuid(4)) // same as above, but explicit
  name String
}

model User {
  id   String @​id @​default(uuid(7)) // will use UUIDv7 instead of UUIDv4
  name String
}

Bug squashing

We’ve squashed a number of bugs this release, special thanks to everyone who helped us! A few select highlights are:

• SQLite db will now be created and read from the correct location when using prismaSchemaFolder.
• Empty Json[] fields will now return [] instead of null when accessed through a join using the relationJoins Preview feature.

Fixes and improvements

Prisma

• Support UUID v7

Language tools (e.g. VS Code)

• Support fetching references for a model

Credits

Huge thanks to @​mcuelenaere, @​pagewang0, @​Druue, @​key-moon, @​Jolg42, @​pranayat, @​ospfranco, @​yubrot, @​skyzh, @​haaawk for helping!

v5.17.0

Compare Source

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

Highlights

VSCode extension improvements

We’re happy to introduce some cool new features that will make your experience with the Prisma VSCode extension even better!

Find references across schema files

The ability to hop between references of a given symbol is really useful in application code and now with the introduction of multi-file schema, we think it’s the perfect time to bring this feature to the VSCode extension!

With the 5.17.0 release, you’ll now have the ability to use the native “find references” feature to find any usage of a given symbol

Added context on hover

When hovering over a symbol that references a view, type, enum, or any other block with multiple values, you’ll now see a handy pop out that shows what is in that block at a glance.

Additional quick fixes

We’ve taken some fixes made by the prisma format cli command and made them quick fixes available to the VSCode Extension. Now, when you have forget a back relation or relation scalar field, you’ll now see in real time what is wrong and have the option to fix it via the extension.

QueryRaw performance improvements

We’ve changed the response format of queryRaw to decrease its average size which reduces serialization CPU overhead.

When querying large data sets, we expect you to see improved memory usage and up to 2x performance improvements.

Fixes and improvements

Prisma Client

• Remove or change This is the 10th instance of Prisma Client being started. Make sure this is intentional. warning

Prisma

• Prisma generate option --allow-no-models doesn't work.

Language tools (e.g. VS Code)

• LSP Rename incorrectly maps field name and appends character

Credits

Huge thanks to @​key-moon, @​pranayat, @​yubrot, @​skyzh for helping!

v5.16.2

Compare Source

Today, we are issuing the 5.16.2 patch release to fix an issue in Prisma client.

Fix in Prisma Client

• nextjs app deployed to vercel edge can't import prisma WASM modul

v5.16.1

Compare Source

Today, we are issuing the 5.16.1 patch release to fix an issue in Prisma client.

Fix in Prisma Client

• dotenv loading issue with PrismaClient
• Prisma Seed Script Fails After Upgrading to v5.16.0 (DATABASE_URL Error)

v5.16.0

Compare Source

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

Highlights

Omit model fields globally

With Prisma ORM 5.16.0 we’re more than happy to announce that we’re expanding the omitApi Preview feature to also include the ability to omit fields globally.

When the Preview feature is enabled, you’re able to define fields to omit when instantiating Prisma Client.

const prisma = new PrismaClient({
  omit: {
    user: {
      // make sure that password is never queried.
      password: true,
    },
  },
});

You’re also able to omit fields from multiple models and multiple fields from the same model

const prisma = new PrismaClient({
  omit: {
    user: { 
      // make sure that password and internalId are never queried.
      password: true,
      internalId: true,
    },
    post: {
      secretkey: true,
    },
  },
});

With both local and global omit, you now have the flexibility to completely remove sensitive fields while also tailoring individual queries. If you need the ability to generally omit a field except in a specific query, you can also overwrite a global omit locally

const prisma = new PrismaClient({
  omit: {
    user: { 
      // password is omitted globally.
      password: true,
    },
  },
});

const userWithPassword = await prisma.user.findUnique({
  omit: { password: false }, // omit now false, so password is returned
  where: { id: 1 },
});

Changes to prismaSchemaFolder

In 5.15.0 we released the prismaSchemaFolder Preview feature, allowing you to create multiple Prisma Schema files in a prisma/schema directory. We’ve gotten a lot of great feedback and are really excited with how the community has been using the feature.

To continue improving our multi-file schema support, we have a few breaking changes to the prismaSchemaFolder feature:

• When using relative paths in Prisma Schema files with the prismaSchemaFolder feature, a path is now relative to the file it is defined in rather than relative to the prisma/schema folder. This means that if you have a generator block in /project/prisma/schema/config/generator.prisma with an output of ./foo the output will be resolved to /project/prisma/schema/config/foo rather than /project/prisma/foo. The path to a SQLite file will be resolved in the same manner.
• We realized that during migration many people would have prisma/schema as well as prisma/schema.prisma. Our initial implementation looked for a .prisma file first and would ignore the schema folder if it exists. This is now an error.

Changes to fullTextSearch

In order to improve our full-text search implementation we have made a breaking change to the fullTextSearch Preview feature.

Previously, when the feature was enabled we updated the <Model>OrderByWithRelationInput TypeScript type with the <Model>OrderByWithRelationAndSearchRelevanceInput type. However, we have noted that there are no cases where relational ordering is needed but search relevance is not. Thus, we have decided to remove the <Model>OrderByWithRelationAndSearchRelevanceInput naming and only use the <Model>OrderByWithRelationInput naming.

Fixes and improvements

Prisma

• Wrong Parameterized Types Sent for SQL Server Queries
• Prisma has no exported member named OrderByWithRelationInput. Did you mean OrderByWithAggregationInput?
• [Driver Adapters]: missing provider compatibility validation
• Disable "Start using Prisma Client" hint logs on prisma generate
• Deploying prisma to CloudFlare pages using Nuxt/Nitro and node-postgres (pg) is using the wrong(vercel) wasm path
• @prisma/adapter-pg modifies node-postgres global type parsers
• @​prisma/adapter-d1 is failing with an import error when called inside vitest tests
• db pull fails with [libs\user-facing-errors\src\quaint.rs:136:18] internal error: entered unreachable code on invalid credentials

Language tools (e.g. VS Code)

• Make prisma-fmt logs to work with language server
• Spans and positions get shifted out of sync when schema includes multibyte characters
• VSCode extension panics when opening an empty prisma schema

Prisma Engines

• [DA] Planetscale engine tests: one2m_mix_required_writable_readable
• [DA] Planetscale engine tests: apply_number_ops

Credits

Huge thanks to @​key-moon, @​pranayat, @​yubrot, @​skyzh, @​brian-dlee, @​mydea, @​nickcarnival, @​eruditmorina, @​nzakas, @​gutyerrez, @​avallete, @​ceddy4395, @​Kayoshi-dev, @​yehonatanz for helping!

v5.15.1

Compare Source

Today, we are issuing the 5.15.1 patch release.

Fixes in Prisma Client

• internal error: entered unreachable code
• Got error 'internal error: entered unreachable code' when trying to perform an upsert.
• Prisma Client errors on SQLite with internal error: entered unreachable code when running 2 concurrent upsert
• ConnectionError(Timed out during query execution.) during seeding
• SQLite timeouts after upgrade from prisma 2 to prisma 4
• ConnectionError(Timed out during query execution.) error when using Promise.all for SQLite
• Improve the error when SQLite database file is locked
• sqlite timeout error multiple queries run one after another
• SQLite times out during query execution when using Promise.all() / concurrent
• internal error: entered unreachable code

v5.15.0

Compare Source

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 d