[kotlin-spring] allOf discriminator children get default discriminator value

[rvlasveld]

What

A child of a discriminator parent already knows its own discriminating value, but kotlin-spring generated that property as an ordinary required constructor parameter. Callers had to pass it on every construction, and nothing stopped them passing a wrong one:

// before
Dog(petType = "Cat", ...)   // compiles, serialises as a Cat

The discriminator property now carries a default taken from the discriminator mapping:

// after
data class Dog(
    @get:JsonProperty("petType", required = true) override val petType: kotlin.String = "Dog",
    ...
)

The oneOf path already did this. This PR generalises that loop to every discriminator parent, so allOf parents are handled by the same code rather than a parallel implementation.

The emitted value is the mapping key, not the schema name

Worth highlighting, because it is easy to get wrong. Given:

discriminator:
  propertyName: className
  mapping:
    DOG: '#/components/schemas/Dog'

the generated default is "DOG", not "Dog". This falls out of CodegenDiscriminator.MappedModel ordering and the de-duplication in DefaultCodegen.createDiscriminator() — subtle enough that it is now pinned by a dedicated test (testAllOfDiscriminatorMappingKeyDiffersFromSchemaName), using the existing petstore-with-fake-endpoints-for-testing-with-cookie.yaml spec which maps DOG/CAT onto Dog/Cat.

Behaviour change

The discriminator property gains a default value, so it becomes optional at the call site. This is source-compatible for existing callers that still pass it explicitly. Callers who passed a wrong value keep compiling — the default only changes what happens when the argument is omitted.

The rewritten loop is also no longer gated on x-is-one-of-interface: it now runs for every model with a discriminator and a non-empty mapped-model set. In practice that widens coverage from oneOf to allOf (the point of the PR) and to anyOf, which shares the same mapped-model machinery.

Tests

KotlinSpringServerCodegenTest: 286 passing, 3 of them new (default from schema name, default from explicit mapping, mapping key ≠ schema name).

Samples

Regenerated all kotlin-spring sample configs. Three samples change — note kotlin-springboot-request-cookie is included, which the earlier revision of this PR had missed:

• kotlin-springboot-x-kotlin-implements — Cat/Dog
• kotlin-springboot-include-http-request-context-delegate — Cat/Dog
• kotlin-springboot-request-cookie — Cat/Dog (shows = "DOG" / = "CAT")

For maintainers

• Independent PR. No ordering relationship with [kotlin-spring] seal allOf discriminator parent interfaces by default #23953; either merge order works. The two touch
  different regions of KotlinSpringServerCodegen.java and different sample files, so the only
  overlap is a both-added resolution at the tail of KotlinSpringServerCodegenTest.java.
• Rebased on master after [kotlin-spring] fix multi-level allOf inheritance (fixes #18206) #23954 merged. That PR added the allOf-multilevel sample, whose
  Animal children are allOf discriminator children — so this PR now also gives Dog, Cat and
  BigDog their className defaults in that sample. This is the regeneration [kotlin-spring] fix multi-level allOf inheritance (fixes #18206) #23954 flagged as
  due once this landed, not a new behaviour.
• Should this be opt-in? It ships on by default with no flag. That seems right — emitting a value the schema already fixes is a bug fix, not a preference — but say the word and I will put it behind an option.
• anyOf is now in scope as described above. It is untested here because no kotlin-spring sample exercises an anyOf discriminator; happy to add one if you would rather have it covered explicitly.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VBn7VWUxZL6mTUbg21xNjm

[cubic-dev-ai]

No issues found across 6 files

_{Re-trigger cubic}

[cubic-dev-ai]

No issues found across 6 files

_{Re-trigger cubic}

[rvlasveld]

@wing328, could you help get this reviewed?

[wing328]

@rvlasveld thanks for the PR

can you please PM me via Slack to go through your PRs and get them merged?

https://join.slack.com/t/openapi-generator/shared_invite/zt-36ucx4ybl-jYrN6euoYn6zxXNZdldoZA

[wing328]

please resolve the merge conflicts when you've time. will try to review, test it before the upcoming stable release (due later this coming week)