@gql.tada/[email protected]

Patch Changes

• Add generate-schema command which takes a URL | path to a JSON file and outputs a graphql schema. Example: gql-tada generate-schema https://example.com ./schema.graphql --header 'authorization: bearer token'
  Submitted by @JoviDeCroock (See #120)

[email protected]

Patch Changes

• Refactor internal GraphQL document parser to use a tokenizer phase, which further utilizes TypeScript’s tail recursion optimization. This should help to further improve type inference performance
  Submitted by @kitten (See #111)

[email protected]

Minor Changes

• Add CLI entrypoint gql-tada capable of generating the types file without the LSP running
  Submitted by @JoviDeCroock (See #58)
• Add graphql.persisted() to create an API for persisted documents that omits a query’s definitions from the output bundle
  Submitted by @kitten (See #98)

Patch Changes

• Allow graphql.scalar to resolve types for input objects
  Submitted by @kitten (See #97)
• Address performance cliff for getDocumentNode inference and object-flattening utilities
  Submitted by @kitten (See #107)
• Refactor type unwrapping for NON_NULL field types (with @_optional and @_required), input types, and variable types
  Submitted by @kitten (See #104)
• ⚠️ Fix $tada not being exported, which can cause projects with isolatedModules: true set from building
  Submitted by @kitten (See #99)
• ⚠️ Fix schema pathname resolution in CLI
  Submitted by @wyattades (See #82)
• Add disableMasking flag to allow fragment masking to be disabled. When this is set to true on the setupSchema interface, fragments won’t be masked, which imitates the behaviour you’d see when adding @_unmask to every single one of your fragments. This is currently considered a preview feature
  Submitted by @kitten (See #69)
• Keep the possible types that are iterated through narrow through repeated abstract type fragment spreads, and provide an optional __typename?: PossibleType field by default so the type checker has an exact property to merge types on
  Submitted by @kitten (See #102)
• Handle inference of input object fields with missing defaultValue properties in introspection
  Submitted by @llllvvuu (See #101)
• Add missing support for input object fields with default values. Previously, input object fields with default values were still marked as required in variables
  Submitted by @kitten (See #73)
• Refactor several internal utility types
  Submitted by @kitten (See #68)
• Updated dependencies (See #58)
  • @gql.tada/[email protected]

@gql.tada/[email protected]

Minor Changes

• Add CLI entrypoint gql-tada capable of generating the types file without the LSP running
  Submitted by @JoviDeCroock (See #58)

[email protected]

Patch Changes

• Remove type name constraint from graphql.scalar’s type name to improve type checking performance
  Submitted by @kitten (See #53)
• Improve performance of several smaller types (Thank you, @deathemperor & @HaiNNT)
  Submitted by @kitten (See #51)

[email protected]

Minor Changes

• Add maskFragments to cast data to fragment masks of a given set of fragments
  Submitted by @kitten (See #43)
• Add graphql.scalar() utility to retrieve or type check the type of scalars and enums
  Submitted by @kitten (See #45)
• Add unsafe_readResult to unsafely cast data to the result data of a given document
  Submitted by @kitten (See #43)

Patch Changes

• Tighten up type strictness to not accept operation documents where fragment documents are expected
  Submitted by @kitten (See #41)

[email protected]

Minor Changes

• Support @_optional and @_required directives on fields overriding the field types.
  When used, @_required can turn a nullable type into a non-nullable, and @_optional
  can turn non-nullable fields into nullable ones. (See “Client-Controlled Nullability” in Graphcache for an example of a client implementing this.)
  Submitted by @kitten (See #32)
• Add support for @_unmask directive on fragments causing the fragment type to not be masked. FragmentOf<> will return the full result type of fragments when they’re annotated with @_unmask and spreading these unmasked fragments into parent documents will use their full type
  Submitted by @kitten (See #31)

Patch Changes

• Format TadaDocumentNode output’s third generic differently. The output of fragment definitions will now be more readable (e.g. { fragment: 'Name', on: 'Type', masked: true })
  Submitted by @kitten (See #31)
• Improve performance of selection and variables inference
  Submitted by @kitten (See #35)
• Improve performance of GraphQL document parser
  Submitted by @kitten (See #34)

[email protected]

Patch Changes

• Prevent type inference for schemas with “huge” root types (i.e. types with an excessive amount of fields) from failing introspection mapping
  Submitted by @kitten (See #25)
• Remove redundant constraint on IntrospectionQuery data. When the full type is used as an extends, the input type (which can be a huge schema), is checked against this type, which forces a full evaluation. This means that TypeScript may spend multiple seconds in recursiveTypeRelatedTo. This work has been eliminated and should help performance
  Submitted by @kitten (See #26)

[email protected]

Patch Changes

• ⚠️ Fix readFragment() not inferring the types of complex fragments, i.e. fragments that derive with a union type
  Submitted by @kitten (See #15)
• Make $tada.fragmentRefs property required. Previously, this was optional (to mirror what GCG’s client-preset does). However, this can lead to invalid checks in readFragment, as it would be able to match types that don’t actually match the fragment refs
  Submitted by @kitten (See #18)

[email protected]

Patch Changes

• ⚠️ Fix __typename literal string not being exact and instead a union of possible types, when the __typename field is put onto an abstract type’s selection set
  Submitted by @kitten (See #11)