0.2.4

Fixed

• Config keys now have more validation and give good error messages - #179
• Code generation now takes the package name from any existing files in the directory - #179
• A bug when auto casting between types in external packages - #180

0.2.3

Added

• support for forcing resolver generation, even when there is a matching field to bind to - #173
• search for config in more places - #176
• docs for config config - https://gqlgen.com/config/

0.2.2

Deprecated

• typemap.json has been deprecated in favour of .gqlgen.yml - #163

Fixed

• Bug when using multiple gopaths - #151
• Generate enums even if there are no models - #159
• Circular references causing generation not to run - #166

0.2.1

Added

• Support for smaller resolver interfaces that don't use _ as a separator, see #106 and #134

Fixed

• generated file header format, gofmt will now ignore these files #139
• missing , when unpacking defaults in input objects #146
• when importing a package use the real package name #147

0.2.0

Added

• Going forward we will start tracking change logs 🎉

Changes

Previously

type User {
    id Int!
    notes [Notes!]!
}

would generate only generate the ID, not the notes:

type User struct {
    id int
}

It now will generate exactly what is in the schema:

type User {
    id int
    notes []Notes
}

This feature was generating a lot of confusion with some expecting the notes field to be present. If you were relying on this behavior you should copy the old model out of the generated code and add it to your types.json.

A future release will add better support for explicitly configuring this behaviour. Proposed syntax is:

type User {
    id Int!
    notes [Notes!]! @resolver
}