nil
nil
nil
- 513 Deprecate returning enumerators from
build_enumeratorthat are not wrapped withenumerator_builder.wrap. The built-in enumerator builders now always wrap.
- 340 Add
cursor.iterationinstrumentation for the query to fetch the next batch of records for the Active Record cursor. - 523 Add interruption adapter for aws-activejob-sqs.
- 515 Fix size of array enumerators.
- 509 - Added CSV batching functionality to
EnumeratorBuilderwithbuild_csv_enumerator_on_batchesmethod andcsv_on_batchesalias. - 512 - Added support for custom timezones on
ActiveRecordEnumeratorandActiveRecordBatchEnumerator. This allows for using cursors with datetime columns whereActiveRecord.default_timezoneis set to:localand Active Record is not using the same timezone as the database.
- 464 - Add interruption adapter for GoodJob.
- 505 - Add interruption adapter for Solid Queue.
- 437 - Use minimum between per-class
job_iteration_max_job_runtimeandJobIteration.max_job_runtime, instead of enforcing only setting decreasing values. Because it is possible to change the global or parent values after setting the value on a class, it is not possible to truly enforce the decreasing value constraint. Instead, we now use the minimum between the global value and per-class value. This is considered a non-breaking change, as it should not break any existing code, it only removes the constraint on new classes. - 443 - Use Sidekiq
:quitcallback to detect graceful shutdown. This makes job-iteration compatible with Sidekiq run in embedded mode. - 445 - Add the
around_iteratecallback, which runs around each call ofeach_iteration. This adds extensibility to build some generic handlers, such as metrics collection and logging. - 450 - Infer which interruption adapter to use from the queue adapter of the job. This deprecates setting
JobIteration.interruption_adapter = <callable>, in favor ofJobIteration.register_interruption_adapter(<queue adapter name>, <callable>).JobIteration.interruption_adapterwill be removed in a future release.
- 437 - Defer reading
JobIteration.max_job_runtimeuntil runtime, instead of closing around the value at the time of job definition. - 431 - Use
#id_valueinstead ofsend(:id)when generating position for cursor based on:idcolumn (Rails 7.1 and above, where composite primary models are now supported). This ensures we grab the value of the id column, rather than a potentially composite primary key value. - 456 - Use Arel to generate SQL that's type compatible for the
cursor pagination conditionals in ActiveRecord cursor. Previously, the cursor would coerce numeric ids to a string value
(e.g.:
... AND id > '1')
- 427 - Use the Rails application logger. Changes from 338 resulted in logging to the original value of ActiveJob.logger, not the one configured by the Rails application.
- 338 - All logs are now
ActiveSupport::Notificationsevents and logged usingActiveSupport::LogSubscriberto allow customization. Events now always include thecursor_positiontag. - 418 - Return
nilfromIteration#perform, to signal not to rely on return value.
- 240 - Allow setting inheritable per-job
job_iteration_max_job_runtime - 310 - Support nested iteration
- 341 - Add
JobIteration.default_retry_backoff, which sets a default delay when jobs are re-enqueued after being interrupted. Defaults tonil, meaning no delay, which matches the current behaviour. - 365 - Support composite primary key as a cursor
- 289 - Fix uninitialized constant error when raising
ConditionNotSupportedErrorfromActiveRecordBatchEnumerator - 346 - Include failed jobs in
total_time - 417 - Ensure that numerical values are deserialized as such and not as strings.
- 190 - Fix updating
times_interruptedandtotal_timewhen job is throttled
- 183 - Add
JobIteration::EnumeratorBuilder#build_csv_enumerator
- 174 - Fix Ruby 3.2 compatibility
- 153 - Re-enqueue jobs only after shutdown hooks have run
- 148 - Revert "Do not evaluate enumerator when throttled", due to backwards incompatibility.
- 87 - Do not evaluate enumerator when throttled (REVERTED)
- 133 - Moves attributes out of JobIteration::Iteration included block
- 107 - Remove broken links from README
- 108 - Drop support for ruby 2.5
- 110 - Update rubocop TargetRubyVersion
- 84 - Call adjust_total_time before running on_complete callbacks
- 94 - Remove unnecessary break
- 95 - ActiveRecordBatchEnumerator#each should rewind at the end
- 97 - Batch enumerator size returns the number of batches, not records
- 91 - Add enumerator yielding batches as Active Record Relations
- 77 - Defer enforce cursor be serializable until 2.0.0
- 73 - Enforce cursor be serializable This is reverted in 1.1.12 as it breaks behaviour in some apps.
- 69 - Fix memory leak in ActiveRecordCursor
- 61 - Call
superinmethod_added
- Preserve ruby2_keywords tags in arguments on Ruby 2.7
- 54 - Fix warnings on Ruby 2.7
- 49 - Log when enumerator has nothing to iterate
- 52 - Fix CSVEnumerator cursor to properly remove already processed rows
- 47 - Optional
sorbet-runtimesupport forJobIteration::Iterationinterface validation
- 45 - Add Throttle enumerator
- 36 - Check method validation at job initialization step
- 36 - Fix CsvEnumerator for Ruby 2.6.3
- 36 - Add case for using default keyword arguments for cursor in #build_enumerator
- 35 - Raise exception if malformed arguments are use in #build_enumerator
It’s been in production at Shopify since 2017. It has support for Rails 5 and 6 🎉
- 34 - remove supports_interruption?
- 30 - Better #each_iteration argument names
- 27 - iteration: don't allow double-retrying a job
- 23 - Remove upperbound constraint on ActiveJob