Skip to content

[pull] master from graphql-java-kickstart:master - #3

Open
pull[bot] wants to merge 548 commits into
lovejavaee:masterfrom
graphql-java-kickstart:master
Open

pull[bot] wants to merge 548 commits into
lovejavaee:masterfrom
graphql-java-kickstart:master

Conversation

@pull

@pull pull Bot commented May 1, 2023 •

Copy link
Copy Markdown

See Commits and Changes for more details.


Created by pull[bot]

Can you help keep this open source service alive? 💖 Please sponsor : )

@cr-gpt

cr-gpt Bot commented May 1, 2023

Copy link
Copy Markdown

Seems you are using me but didn't get OPENAI_API_KEY seted in Variables/Secrets for this repo. you could follow readme for more information

@pull pull Bot added the ⤵️ pull label May 1, 2023
oryan-block and others added 28 commits October 17, 2026 12:07
…puts

Since #842, a scalar used only in a Map input is registered, so the Map test no longer needs another input to reach its scalars. The typed cases share one schema and resolver, and the UUID scalar comes from the end-to-end fixtures.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The interface test's Dog now extends a sealed Mammal and implements the
sealed Animal directly, so it's found twice among Animal's subclasses and
the test fails without the distinctBy. The lookup is renamed to
sealedSubclassMatches and the lists of matches to sealedMatches and
matches, so they don't read like sealedSubclasses, which returns classes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…lver method

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An undefined type that only has an object or input extension is now reported as not defined. The directives test also covers a non-repeatable directive that only an extension applies, and checks that each directive is wired once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…or enum

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The classes were remembered per argument, so after one request with a value that couldn't be converted, later values of the same class were embedded even when they converted. A deserializer that reads the serialized form then got nothing, and the same request gave a different result. The memo now lasts one convert call. The guard test checks the same request before and after such a value, with a tree-reading deserializer in place of the GeoPoint one.

The writer is built once and the reader on first use, instead of on every read.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Let input types share a dictionary class
Find union members and interface implementors among sealed subclasses
# Conflicts:
#	src/main/kotlin/graphql/kickstart/tools/SchemaClassScanner.kt
…tensions

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Pass coerced custom scalar values through to input objects and Map arguments
Merge type extensions into the types they extend
… the proxy's interfaces

Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Claude <noreply@anthropic.com>
Warn at build when a resolver can't be called on its JDK dynamic proxy
A misspelled property on an input class, like `technolgies` for
`technologies` in #77, only showed up once a client sent the field:
Jackson then failed to convert the argument, or dropped the value if the
ObjectMapper ignores unknown properties. The scanner's getter lookup
can't tell, since Jackson also maps setters, creators, naming
strategies, aliases and any setters.

Now the SchemaParser asks Jackson. For each resolver argument whose type
is an input object, it looks up the deserializer that the field's own
ObjectMapper converts the argument with, and checks that it has a
property for every field of the built input object type, through lists,
arrays, Optionals and nested input objects. A field it doesn't have is
read into a placeholder object with unknown properties made to fail, so
ignored names, any setters and ignoreUnknown count as mapped without
running the class's constructors or setters.

- New SchemaParserOptions.unmappedInputFields: WARN by default, FAIL
  throws one SchemaError listing every field, IGNORE skips the check
- MethodFieldResolver keeps its ObjectMapper, so the check and the
  conversion use the same one
- The scanner's "Cannot find a public getter or field" warning is now
  debug, since it fired for setter-only classes and similar

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A List, array or Optional of a class with @JsonTypeInfo got the base
class's deserializer from the container, so the type id and every field
that only a subclass has were reported as unmapped, which failed the
build in FAIL mode for inputs that convert fine. The check now looks up
the elements' type info the way Jackson does when it creates the
container's deserializer, from the property's annotation or else from
the class or default typing, and skips them like a polymorphic property.

- An input type used by fields whose ObjectMappers handle unknown
  properties differently is reported once for each consequence, so a
  field that fails isn't hidden behind one that drops the value
- Errors such as NoClassDefFoundError while looking up a deserializer
  are logged at debug like exceptions, since the lookup used to run on
  the first query and mustn't fail a build that used to succeed
- Tests for Map<String, bean> arguments and polymorphic properties, and
  contains assertions that show the actual message

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
oryan-block and others added 30 commits October 23, 2026 22:38
…maven-setup

Use Kotlin's documented Maven setup to compile Kotlin before Java
setup-java v6.1.0 writes repositoryOrigins into settings.xml, so the snapshot deploy no longer needs maven.repository.credentialScope=id to send credentials under Maven 3.10. The release deploy gets the same origin, which removes its credentials warning.
…itory-origins

Declare central.sonatype.com as a repository origin
spring-aop is a provided dependency, but module-info required it, so an
application on the module path without Spring failed at startup with
'Module spring.aop not found'.
The Renovate rule kept Spring below 6 because it needed Java 17, which
is now the minimum.
…ppet

Gradle 7 removed the compile configuration.
Nothing uses cglib-nodep or objenesis. The release option already sets
the source and target, so it now reads java.version too. The
surefire-junit4 provider stays, with a comment on why.
Jackson skips FAIL_ON_EMPTY_BEANS for a class with annotations and writes it as {}. Every Kotlin class has @metadata, so a Kotlin scalar class without visible properties was read back as a new default instance, or as {} into an ObjectNode. The argument converter now fails on any bean written without properties, so the value is embedded like an unannotated Java class.
A custom scalar without visible properties is now embedded, and Map and JsonNode deserializers reject embedded values, so a Kotlin scalar class in a Map or ObjectNode field failed where it used to convert through {}.
…-empty-scalars

Keep the values of Kotlin custom scalars without visible properties in arguments
…empty-scalars

Fail on custom scalars without visible properties written with a type or object id
…-link

Link the How graphql-java-tools builds a schema wiki page from the README
Optional<DataFetcherResult<T>> bound T's type to the DataFetcherResult class, so graphql-java got the DataFetcherResult as the field value. DataFetcherResult and Optional are now peeled in either order, each once per value, and the Optional is unwrapped before graphql-java sees the DataFetcherResult. A DataFetcherResult or an Optional twice in one value, such as Optional<DataFetcherResult<Optional<T>>>, now fails the build.
…ata-fetcher-result

Unwrap a DataFetcherResult inside an Optional
GenericWrapper.withCompletionStageTransformer registers a wrapper whose
transformer returns a future of its content, like Mono::toFuture. The
unwrapper then waits on that future before the declared one, so
Mono<CompletableFuture<Item>> resolves to the Item. Wrappers registered
with withTransformer keep handing their result to a declared future as is.
…-returning-wrapper

Wait on a future inside a generic wrapper whose transformer returns a future
…r-patch

Update all non-major dependencies
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants