Skip to content

Latest commit

 

History

History
38 lines (27 loc) · 3.33 KB

UPGRADE.md

File metadata and controls

38 lines (27 loc) · 3.33 KB
Raw

Upgrades

This document highlights breaking changes in releases that will require some migration effort in your project. As we move towards a 1.0.0 release these will be restricted to major upgrades only, but currently, whilst the API is still being fleshed out in the 0.x releases, they may be more frequent.

0.5.x -> 0.6.0

  • The test.NewConfig() helper function has been removed. Instead, just use config.NewBuilder() to build your config in your test code.

0.4.x -> 0.5.0

  • The DB field in config.Config has now been renamed to db and is therefore unavailable outside of the config package
  • The data.MigrateDatabase() function signature has changed: it now no longer needs a *config.Config value as its second argument, instead it takes the schema name as a string instead (this should not affect user-land code).
  • The data.NewDB() function signature has changed: it now no longer needs a log.Logger value as its second argument.

0.3.x -> 0.4.0

  • The handler signature has now changed. The first argument is now a context.Context value to allow handlers to act upon cancellations and timeouts from this module.
  • config.NewConfig() function has been removed in favour of using the config builder, via config.NewBuilder(). See the docs for more information on how to use this.
  • test.NewConfig() argument types have changed to align it with the builder that it now uses underneath the hood:
    • The sourceTopics argument is now a []string instead of string
    • The retryIntervals argument is now a []int instead of string
  • config.Builder.SetTopicNameDecorator() has been renamed to config.Builder.SetTopicNameGenerator() which matches the concept and underlying field names better

0.2.x -> 0.3.0

  • test.ConsumeFromKafkaUntil() now returns an error if the consuming failed, you will need to handle it if you are using this test helper in your integration test suite
  • consumer.Start()'s parameters have now changed, there is no need to provide a channel for failures as this is now managed internally
  • consumer.Start() now returns an error if the consuming failed, you will need to handle this error, likely by logging it and exiting
  • consumer.NewCollection() has been renamed to consumer.newKafkaConsumerCollection(), and is now no longer part of the public API of this module
  • The consumer.ConsumerCollection interface has been renamed to consumer.collection, and is now no longer part of the public API of this module
  • The Failure struct type has been moved to data/failure/model package, previously it was in the module root
  • consumer.NewConsumer() function has been renamed to consumer.newConsumer(), and is now no longer part of the public API of this module.

0.1.x -> 0.2.0

  • config.NewConfig() now returns an error value, as well as a *config.Config value, so you will need to check for an error and handle it accordingly
  • You can now use test.NewConfig() to create a config.Config value for testing. See testing for more information.
  • You can now use the test.ConsumeFromKafkaUntil() function in your integration tests to make it easier to wait until messages are consumed and conditions have been satisfied. See testing for more information.