v0.7.1

Added

• implement sibling task spawning with automatic parent-ID inheritance (#89)

v0.7.0

Added

• implement observability metrics export (#20) (#88)
• [breaking] implement priority aging and weighted fair scheduling (#84)
• [breaking] implement token-bucket rate limiting per task type and group (#82)
• [breaking] implement group-level pause and resume (#81)
• add Throughput to all criterion benchmarks so critcmp shows ops/sec (#78)
• [breaking] normalize timestamps from TEXT to epoch millisecond INTEGER (#75)
• expose fail_fast() on SubmitBuilder and DomainSubmitBuilder (#73)
• add tag key prefix queries for namespace-scoped discovery (#71)
• [breaking] support passing state from execute() to finalize() via typed Memo (#69)

Fixed

• use distinct critcmp group names so PR benchmark diff shows two columns (#77)
• use distinct critcmp group names for PR benchmark comparison (#74)
• use distinct critcmp group names for PR benchmark comparison (#72)

Other

• add migration guide for 0.7.0 (#85)
• consolidate paused_groups migration into standard include_str! pattern (#83)
• [breaking] improve benchmark throughput across submit, dispatch, retry, and failure paths (#80)
• skip transaction for retry requeue and tune slow benchmarks (#79)
• [breaking] consolidate migrations from 9 chronological files into 4 object-oriented files (#76)

v0.6.0

Added

• [breaking] replace &TaskContext with DomainTaskContext<D> for type-safe child spawning (#68)

Other

• migrate all tests/benches from TaskExecutor to TypedExecutor (#66)

v0.5.3

Other

• batch dispatch and completion SQL to reduce round-trips (~56% faster) (#61)

v0.5.2

Other

• reduce SQL round-trips and CPU overhead in scheduler hot paths (#60)
• coalesce task completions into batched transactions (#59)
• reduce SQL round-trips in scheduler hot paths (#57)

v0.5.1

Fixed

• (bench) eliminate per-sample scheduler setup cost in history benchmarks (#55)
• (bench) remove premature cancellation token call in history benchmark setup (#54)
• (ci) bootstrap _benchmarks branch on first push to main (#53)
• (ci) restore stderr capture for benchmark output on main (#51)
• (ci) exclude lib target from cargo bench to fix benchmark CI (#49)

Other

• decompose internal god objects into focused, single-responsibility modules (#56)
• eliminate stringly-typed history status and DRY violations (#52)

v0.5.0

Added

• [breaking] domain-centric API with typed event streams (#48)

Fixed

• (ci) capture benchmark output from stderr in workflow (#47)

Other

• add multi-module guides, benchmarks, and CI benchmark workflow (#45)

v0.4.0

Added

• implement multi-module API (#44)
• adaptive retry with configurable backoff strategies (#42)
• add task metadata tags (#40)
• task dependencies with blocked status, cycle detection, and failure cascading (#39)
• delayed and recurring task scheduling (#38)
• task TTL with automatic expiry, sweep, and child inheritance (#33)
• task superseding with atomic cancel-and-replace (#32)
• task cancellation with abort hooks (#31)
• bulk task submission with BatchOutcome, BatchSubmission builder, intra-batch dedup, and chunking (#30)
• add byte-level progress reporting with EWMA throughput tracking (#29)

Other

• split large store modules into focused sub-modules (#43)
• rewrite documentation to be user-facing (#28)
• [breaking] consolidate IO fields into IoBudget and introduce TaskEventHeader (#26)

v0.3.1

Fixed

• scheduler performance and correctness improvements (#24)
• (taskmill) atomic parent resolution and weak scheduler reference in TaskContext (#22)

Other

• (taskmill) split large modules into focused submodules and optimize completion hot path (#25)
• (taskmill) add integration tests and criterion benchmarks (#21)

v0.3.0

Added

• add network IO pressure and per-group concurrency limiting (#7)

Other

• [breaking] builder pattern for TaskSubmission and accessor methods for TaskContext (#6)
• [breaking] simplify executor API with incremental IO tracking and expanded docs (#4)