Temporary workaround for Union-Types in the distant future

[medz]

Temporary workaround for Union-Types in the distant future

In the planned full Prisma ORM feature support, we voted to wait for Dart 3 Union-Types support results.

But after months of waiting, the Dart language team is reluctant to announce any news and developments related to union types. My vision:

1. For multi-type input scenarios such as where, the dynamic type is used, and we add a check method inside the function to check that the input is a supported type.
2. The returned joint demand will not be processed temporarily.
3. For incoming parameters, we add temporary docs to describe it.

Disadvantage: It is impossible for users to know what the incoming type is without reading the documentation or source code.
Advantages: Completed ahead of time, when Dart Union-types are supported, we can produce types for them (dynamic is replaced by a specific union type)

[medz]

Alternatively, we construct a nested view type for nesting:

where = PrismaUnion<A, PrismaUnion<B, C>>

This will obviously increase the complexity of the condition. It is worth noting that this method is adopted by the early version 1.0.

[Linloir]

Dart 3 is out alongside Google I/O yesterday! looking forward for updates~

[medz]

Dart 3 is out alongside Google I/O yesterday! looking forward for updates~

No Union-Types and macros, little pity.

[Linloir]

Dart 3 is out alongside Google I/O yesterday! looking forward for updates~

No Union-Types and macros, little pity.

Is it possible to make union structures such as something below?

class UnionNullableStringWhereClause {
  factory string(String val);
  factory null();

  String? value;
  bool isNull;
}

By properly defining factory constructors it's possible to deduce whether the value is a string or a null.

And by using a nullable UnionNullableStringWhereClause type in the where clause it'll be easy to tell whether the user is declaring a null for the string or just left the whole where empty.

[medz]

I made this structure once, and the code size doubled. It doesn't feel like the best solution. Currently I'm thinking about dynamic or nested Union views. After dynamic, Dart 3 will migrate painlessly after the release of Union, while nested attempts will require a lot of code changes for subsequent upgrades.

dynamic Advantages: Painless upgrades, Disadvantages: Developers must read the comments, otherwise they will not know the specific content that needs to be passed in.

[medz]

I haven't waited long for the Dart union types feature. PrismaUnion is built in the new version to implement structure nesting. Please follow #290