1.7.3

• openapi: fix detection of default (undefined) schema implementation
• openapi: better support for annotation instances created directly in Kotlin code
• server/openapi: fix analyzing of class-decorated handlers for OpenAPI generation
• core/openapi: fix initialization of Converter-supported types in OpenAPI generation
• json: deprecate KeyConverter and ValueConverter in favor of the ones in klite-core module
• server: basicAuth() helper
• jdbc: DockerCompose.startDB() will now wait until DB actually accepts connections
• jdbc: support kotlin.uuid.Uuid and kotlin.time.Instant by default

1.7.2

• server: HttpServer now accepts connections inside of workerPool, not single thread - for better scalability under very heavy load
• server: start() method now accepts socketBacklog parameter that can be fine-tuned if you have 1000s of parallel connections
• server/jobs: ensure that JobRunner logs with the same instanceId as the Server (they use the same RequestIdGenerator now)
• server: metrics() can be provided with custom access annotations and includes requestsActive metric now
• server: metrics() can output Prometheus/OpenMetrics format if Accept header requests it
• jdbc: metrics() for PooledDataSource connection pool include "count" and "dropped" metrics now
• jdbc: changesets now support providing multiple comma-separated contexts
• jdbc: PooledDataSource now logs number of connections in the pool when dropping previous ones
• jdbc: BaseCrudRepository.get(forUpdate = true) now uses "for no key update" for Postgres to avoid unnecessary child rows locking

1.7.1

• GitHub organization changed to keksworks - update your dependency organization
• jdbc: PooledDataSource.queryTimeout can be configured separately from pool connection timeout
• server: MultipartParser and MultipartRenderer now support binary data/files
• server: nullable List of @QueryParam will now get null value if parameter is missing, not empty list
• server: introduced Metrics for publishing of app metrics, and a metrics() handler
• server: RequestTransactionHandler now supports exclude methods, to skip transactions for e.g. GET/HEAD requests
• core: new extensions for easier working with Java HttpClient, like get()/post()/put()
• xml: new module for fast and lightweight XML parsing
• Java 25 support

1.7.0

Note: this release contains small backwards-incompatible changes that may need adjusting in some projects, see below:

• core: TSIDGenerator can now be used with custom classes outside TSID companion
• core/jdbc: Any.toValues()/BaseCrudRepository.persister() will now return KProperty1 keys, not Strings for added type-safety.
  • If you need to use String keys, use toValues().mapKeys { it.key.name }
  • You may get "column specified more than once" errors if you have code like entity.toValues() + ("hello" to 123), replace "hello" with Entity::hello in that case
  • It is now recommended to use toValues() for both providing values and skipping them, e.g. entity.toValues(Entity::field1 = "other value, skip = setOf(Entity::field2)) - this avoids creation of intermediate maps and is faster
  • PropValue type alias now accepts 2 type arguments, so replace usages of PropValue<T> with PropValue<T, *>
• json: toJsonValues() introduced for the cases when you want to modify how an entity is serialized into json (e.g. remove some keys)
• jdbc: introduce @Column annotation to override DB column names for entity fields, previous internal typealias Column renamed to ColName
• jdbc: make it possible to override how Enum/Array is stored using JdbcConverter, default to toString() for enums
• jdbc: db.upsert() will now work as SQL MERGE for non-Postgres databases
• server: use<>() call will fail if called inside context {} for Extensions that need Server-level initialization
• server: allow body in DELETE requests, as sometimes people do that #117
• json: improved detection if inline classes should be rendered as their value type or a string
• jackson: DeserializationFeature.FAIL_ON_TRAILING_TOKENS is now enabled by default
• smtp: SmtpEmailSender will now log the successful email sent (to address and subject)
• openapi:
  • skip @AttrParam from OpenAPI spec
  • allow specifying SwaggerUI config parameters in openApi() function
  • skip @apiresponse responseCode=default

1.6.16

• jdbc: restore exactly the same checksums of DBMigrator changesets containing comments

1.6.15

• core: log current active ENV when it is initialized
• jdbc: improve AlreadyExistsException message (always starts with errors.alreadyExists)
• jdbc: better handling of comments inside quoted strings in migration scripts
• csv: support for providing a different encoding than UTF-8 when parsing or generating
• openapi: adding openApi() to context will now add /openapi.json, /openapi.html and /openapi routes, which will also serve Swagger-UI automatically

1.6.14

• jdbc: add helpful details to Postgres exception "no hstore extension installed"
• core: TSID.deterministic introduced for using in test data generation
• json: TSGenerator updates
  • wraps records with enum keys into Partial to avoid TS wanting all the keys to be present
  • outputs provided custom types as separate types and references them
  • accepts -o and -p arguments for simpler usage
  • optionally can generate test data from Kotlin objects using -t

1.6.13

• core: added PATCH method to TypedHttpClient
• jdbc: can use Config ENV vars in migration scripts
• jdbc-test: JUnitAssertionImprover removed, as Atrium 1.3+ improved error messages itself
• server: disallow direct assignments to HttpExchange.responseType to avoid confusion, HttpExchange.send() or startResponse() must be used to bypass renderers

1.6.12

• smtp: new module introduced for sending emails over SMTP
• server: expose Server.listen and bound Server.address as separate properties #93
• server: Server.use() can register extensions that implement multiple supported interfaces at the same time
• server: useOnly() will now add parser/renderer if it wasn't yet registered, to avoid confusion
• server: HttpExchange.rawBody can now be accessed more than once (is a lazy property now)
• jdbc: distinct and notDistinct operators introduced for where expressions
• jdbc: PooledDataSource: DB connections closed by server now correctly decrease connection pool size
• json: fixed some very specific regressions in TSGenerator

1.6.10

• server: Server(InetSocketAddress(0)) can now be used to bind to any available port.
• server: make idle restarts quick #96
• server: initialize HttpExchange.pathParams even no route is matched (404), for decorators #82
• server: put list into queryParams if user specifies multiple parameters with the same name, introduce e.queryList #63
• jdbc: between operator introduced with open and closed ranges, also in and notIn
• jdbc: @NoTransaction can now be used on jobs
• jdbc: fixed usage of multiple different DataSources when there is an active transaction
• jdbc: allow calling of PooledConnection.close() multiple times #80
• jdbc: support for automatic persisting/mapping of arrays of @JvmInline classes (e.g. TSID)
• json: fix TSGenerator on Windows
• json: TSGenerator will now use more type-safe string template types for java.time classes, e.g. ${number}-${number}-${number} instead of string
• slf4j: allow providing of LOGGER_CLASS via .env file if Config.useEnvFile() is called before any logging calls #92
• jackson: serialize enums using their toString() method by default, this fixes openapi module usage with jackson #88
• liquibase: do not close jdbc connection if it was passed by user #81