Migration from Instancio

Use this mapping to move Instancio fixture code to kRandom. The two libraries share the object-population core but differ in philosophy: Instancio is a fixture DSL with selectors and models; kRandom pairs reflection-based population with semantic data and explicit rules. The common cases map directly; the honest-gaps section lists what does not (yet).

Install

dependencies {
    implementation("io.github.frikit:krandom-core:2.10.0")
}

API mapping

Instancio	kRandom equivalent
`Instancio.create(User.class)`	`Generators.ofObject(User.class).generate()`
`Instancio.ofList(User.class).size(10).create()`	`Generators.ofObject(User.class).generateList(10)`
`Instancio.of(User.class).set(field(User::getEmail), v).create()`	`ObjectFaker.of(User.class).ruleFor(User::getEmail, () -> v).generate()`
`.supply(field(User::getId), gen)`	`.ruleFor(User::getId, gen)` (any `Generator<T>` or lambda)
`.ignore(field(User::getPassword))`	`.ignore(User::getPassword)` / `.objectExcludeField("password")`
`.withNullable(field(...))`	`GeneratorConfig.builder().objectNullProbability(0.1)`
`.onComplete(...)`	`ObjectFaker.afterGenerate(consumer)` / `.postProcess(operator)`
`.withSeed(42)`	`GeneratorConfig.builder().seed(42L)`
`.withMaxDepth(3)`	`.objectMaxDepth(3)`
`Instancio.of(Model<T>)`	`ObjectModel<T>.faker()` / `.generate()`; compose models with `.and(...)`
`generate(field(...), gen -> gen.ints().range(1, 10))`	`.ruleFor(User::getCount, Generators.ofInt(1, 10))`
`@ExtendWith(InstancioExtension.class)` + `@Seed(42)`	`@ExtendWith(KrandomExtension.class)` + `@KrandomSeed(42L)` from `krandom-junit` — same failure-seed reporting; see the JUnit Extension guide

Before and after

Instancio:

User user = Instancio.of(User.class)
    .set(field(User::getEmail), "owner@example.test")
    .ignore(field(User::getPassword))
    .withSeed(42)
    .create();

kRandom:

GeneratorConfig config = GeneratorConfig.builder().seed(42L).build();

User user = new ObjectFaker<>(User.class, config)
    .ruleFor(User::getEmail, () -> "owner@example.test")
    .ignore(User::getPassword)
    .generate();

A rule may repeat its value on every object. To request a unique field, add objectUniqueField("email") to the configuration, and the constant rule above then fails on a second object from the same faker.

What kRandom adds over Instancio

Semantic field values — firstName, email, city resolve to realistic domain data, not arbitrary strings (this is the documented throughput-for-realism trade-off; see the README performance section).
Bean Validation awareness — 21 Jakarta constraints (@Email, @Min, @Pattern, @Size, …) are honored during population without extra setup.
Bulk structured export — Field + Schema produce JSONL, JSON, CSV, XML, SQL, YAML, and TOML output, including streaming writers and existing-object projections.
Locale-aware data across 50 supported locale variants (35 native datasets and 15 curated fallbacks).

Honest gaps

Feeds (CSV/JSON-backed data sources): not shipped; on the roadmap.
Selector depth scoping (atDepth(...)): use objectMaxDepth plus per-field rules; per-depth selectors have no direct equivalent.
Selector groups and arbitrary selector predicates: typed paths select one property at a time; broad predicate rules remain configured through GeneratorConfig field predicates.
JPA-specific constraints: Bean Validation is native, but JPA metadata such as @Column(length=...) is not interpreted.
JUnit generated-value injection and parameter sources: kRandom reports and replays per-test seeds, but does not yet provide equivalents to Instancio’s generated field/parameter injection and parameterized source APIs.