Prisma ORM for Dart 3.0

[medz]

Due to the limitation of the Dart language, many functions of Prisma ORM for Dart cannot be realized, the most common ones are select, include.

There are many other problems:

1. Codegen is too complex
2. Difficult to add new features

The guesses for the v3 version are as follows:

1. Give up type safety and use Map
2. Give up building a private client and use Prisma DMMF to build a general client
3. Abandon the binary engine (although it supports conventional platforms and architectures, but it is troublesome for serverless or Docker), and build a brand new engine suitable for Dart.
4. For parameter input, use chain reaction to build

[medz]

For 3.0, there are the following technical requirements:

1. Dart SDK requires 3.0 (better for copying Prisma TS/JS client)
2. Looking for a better JSON serialization build Abandon the build_* ecological package from the project
3. Provide a complete Prisma C API implementation, discarding the binary query engine

[wstrange]

It would be great to see the engine implemented in pure Dart, even if it only means supporting a subset of databases and features (postgres, sqllite?).

The binary engine makes packaging and distribution harder (can't just compile a pure dart server). I suspect it would make debugging easier as well (no binary interface to cross).

[medz]

It would be great to see the engine implemented in pure Dart, even if it only means supporting a subset of databases and features (postgres, sqllite?).

The binary engine makes packaging and distribution harder (can't just compile a pure dart server). I suspect it would make debugging easier as well (no binary interface to cross).

@wstrange Now the problem is, I'm an introductory Rust learner. I tried writing a dynamic engine library to make it behave better with the engine, but I couldn't solve the upstream breaking issue.

prisma/prisma-engines#3757

This issue has already flagged how to fix the upstream breakage, and I think it will be a long time before that happens.

In addition, Dart3's macro support release date is undecided, and Union-Types does not know whether it will appear in Dart 3 within this year, because it is still making proposals.

Under the current limited conditions, I can only do the engine client based on maximizing parameter selection and cannot fully realize all the requirements of Prisma (unless I use Map)