v0.30.dev14

What's Changed

• Switch to a more robust CaptureRuntimeView design by @MOZGIII in #535
• Bump rustls and some of its deps by @MOZGIII in #702
• Macro-powered interpreter composition by @MOZGIII in #536
• Apply useful changes from the cancelled operations refactor by @MOZGIII in #539
• Generalize VM values over flavors and make a python flavor by @MOZGIII in #544
• Make workflow arguments opaque bytes at the protocol level by @MOZGIII in #559
• Improve the RoundTripMetrics composition by @MOZGIII in #560
• Simplify the worker-core and related API surface by @MOZGIII in #561
• Rework worker protocol by @MOZGIII in #562
• Eliminate JSON intermediary conversions for values by @MOZGIII in #563
• Rename the WorkflowArgumentValue to just Value (and the rest accordigly) by @MOZGIII in #564
• Cleaned up the value-related conversion machinery by @MOZGIII in #565
• Rename vm-value-convert-proto to vm-value-python-convert-proto by @MOZGIII in #566
• Split and rework the worker-core APIs at the rust side by @MOZGIII in #567
• Make the action execution loss explicit at the worker API by @MOZGIII in #568
• Distinct error states at the worker-message-protocol by @MOZGIII in #701
• Avoid re-exports of the PendingPromiseError by @MOZGIII in #569
• Adjust and cleanup the conversion API surface by @MOZGIII in #570
• Fix tonic build composition by @MOZGIII in #553
• Explicitly propagate action loss though the action completions reconciler by @MOZGIII in #571
• Make WorkflowOutcome shape explicit at the proto level by @MOZGIII in #572
• Define proper protocol-level boundaries for value trees by @MOZGIII in #573
• Generalize the action-runtime-convert around the actual value type by @MOZGIII in #574
• Remove the webapp for the rework by @MOZGIII in #621
• Metrics to track the workers by @MOZGIII in #622
• Core HTTP API machinery by @MOZGIII in #624
• Add lossy-batcher by @MOZGIII in #625
• Add retention-sweeper by @MOZGIII in #626
• Add sqlx-postgres-schema-pool by @MOZGIII in #627
• Proper node id type by @MOZGIII in #628
• Essential metrics subsystem core parts by @MOZGIII in #629
• Add observability-store-postgres (and -migrations) by @MOZGIII in #630
• Allow configuration of max connections for the main db pool by @MOZGIII in #631
• Observability core and essential-metrics config by @MOZGIII in #632
• Wire the observability core and essential-metrics pipeline through vertically by @MOZGIII in #633
• Add counter-rate by @MOZGIII in #634
• Port the soak harness to essential metrics by @MOZGIII in #635
• Remove the worker-status pipeline in favor of essential metrics by @MOZGIII in #636
• Add http-api-types by @MOZGIII in #659
• Add api-essential-metrics-http by @MOZGIII in #637
• Serve the observability API from start-workers by @MOZGIII in #638
• Serve the HTTP interface from soak runs by @MOZGIII in #639
• Serve the HTTP interface from the example app by @MOZGIII in #640
• Serve Swagger UI from the API by @MOZGIII in #641
• Extend the prometheus seconds buckets down to microseconds by @MOZGIII in #642
• Add node-sequence by @MOZGIII in #643
• Add the observability-events core and payload by @MOZGIII in #644
• Add cursor-core by @MOZGIII in #645
• Add query-limit by @MOZGIII in #656
• Add the observability-events backend traits by @MOZGIII in #646
• Add observability-events-emitter and -compat by @MOZGIII in #647
• Implement observability events backends at observability-store-postgres by @MOZGIII in #648
• Add the page, the cursor and the limit to http-api-types by @MOZGIII in #660
• Add api-observability-events-http-types by @MOZGIII in #661
• Add timestamp crate by @MOZGIII in #703
• Add api-observability-events-http by @MOZGIII in #649
• Wire the observability-events pipeline through vertically by @MOZGIII in #650
• VM driver hooks machinery by @MOZGIII in #651
• Feed the observability events from the VM driver hooks by @MOZGIII in #652
• Route the VM driver run logging through the hooks by @MOZGIII in #653
• Trace every transient VM driver run via the hooks by @MOZGIII in #723
• Per-VM timeline of the observability events by @MOZGIII in #654
• Observability state API and the instance view by @MOZGIII in #655
• Reword the scheduler-core crate doc by @MOZGIII in #657
• Separate read pool for observability database by @MOZGIII in #662
• Set a statement timeout on the observability pools by @MOZGIII in #663
• Sweep the observability retention in chunks by @MOZGIII in #664
• Record the snapshot_persisted events only when asked by @MOZGIII in #665
• Carry the observability events' drops in the essential metrics by @MOZGIII in #724
• Bring the observability subsystem up under the benchmark on request by @MOZGIII in #666
• Take the essential-metrics series range as the timestamp proxy by @MOZGIII in #721
• Take the essential-metrics series range as the timestamp proxy by @MOZGIII in #752
• List the observability knobs in the start-workers configuration doc by @MOZGIII in #731
• Better clarify the WAYMARK_OBSERVABILITY_READ_DATABASE_URL purpose in the docs by @MOZGIII in #753
• Add os-shutdown-requests crate by @MOZGIII in #668
• Fix soak harness Ctrl+C handling by @MOZGIII in #667
• Port start-workers shutdown signals to os-shutdown-requests by @MOZGIII in #671
• Add task-supervisor crate by @MOZGIII in #672
• Supervise every start-workers task from main by @MOZGIII in #673
• Rework the worker pool initialization and error surface by @MOZGIII in #674
• Drop the transient worker pool bringup's caller reminder by @MOZGIII in #728
• Act once per worker generation on recycle reports by @MOZGIII in #720
• Add the managed-spawner crates by @MOZGIII in #675
• Switch the secondary binaries to task-supervisor by @MOZGIII in #676
• Keep the execution driver loop alive while the VM drivers execute by @MOZGIII in #677
• Run the inline worker pool's actions from a pool loop that waits for them by @MOZGIII in #678
• Rework the managed-process API around observing the child's exit by @MOZGIII in #679
• Stop the soak worker gracefully through managed-process by @MOZGIII in #680
• Fix the worker bridge bringup's returned-address doc by @MOZGIII in #681
• Drain secondary binaries' supervisors on every exit by @MOZGIII in #682
• Make workload-pinning-manager's outcome an error by @MOZGIII in #683
• Add an error converter to fn-main-common by @MOZGIII in #684
• Route the binaries' supervisors through managed-spawner-supervised by @MOZGIII in #685
• Make the state sweeper a future at execution-bringup by @MOZGIII in #686
• Bring everything up through the managed spawner by @MOZGIII in #687
• Correct execution-bringup cancellation token application by @MOZGIII in #688
• Bring the Prometheus exporter up through the managed spawner by @MOZGIII in #689
• Bring the tokio metrics reporters up through the managed spawner by @MOZGIII in #690
• Make the transient execution's response stream own its per-request tasks by @MOZGIII in #691
• Document what lossy means for the lossy batcher by @MOZGIII in #692
• Route the batchers' shutdown through the force shutdown token by @MOZGIII in #693
• Make the lossy batchers end on their last handle by @MOZGIII in #694
• Stop the state-manager sweeper loop only when its state is gone by @MOZGIII in #695
• Stop the demand registry bookkeeping and the ackers when their registrars and handles are gone by @MOZGIII in #696
• Report the worker pool shutdown from poll_complete as None instead of an error by @MOZGIII in #697
• Split the worker pool handle into requests and completions sides by @MOZGIII in #698
• Stop the lock renewal loop as soon as its channel is closed and nothing is tracked by @MOZGIII in #699
• End the lock renewal with the worker pool by @MOZGIII in #700
• Add schedules to the example app's dev reset by @MOZGIII in #705
• Declare the typing-extensions dependency by @MOZGIII in #706
• Declare exactly the tonic and tokio features the bridge crates use by @MOZGIII in #707
• Gate rustdoc warnings in the extended Rust lint by @MOZGIII in #708
• Document the server bringups' spawned tasks by @MOZGIII in #709
• Replace RoundTripMetrics with ActionRoundTrip in the worker message protocol by @MOZGIII in #710
• Drop the worker's inert cleanup timeout by @MOZGIII in #711
• Bring the docs in line with the current knobs and helpers by @MOZGIII in #712
• Raise a typed error when a workflow fails by @MOZGIII in #713
• Mark the exception class hierarchy as input to the planned base-class matching by @MOZGIII in #765
• Run the binaries' supervisors on the report's Error form by @MOZGIII in #714
• Add fn-main-lifecycle crate by @MOZGIII in #715
• Run the start-workers lifecycle through fn-main-lifecycle by @MOZGIII in #716
• Take a shutdown future in the worker pool loop by @MOZGIII in #717
• Relax the pinning error bounds at the execution bringup by @MOZGIII in #718
• Run the Rust tests with all features by @MOZGIII in #722
• Bring the AGENTS.md trait-method and metrics rules up to date by @MOZGIII in #729
• Log level tweaks by @MOZGIII in #730
• Mark the soak harness's observability connect for the database setup rework by @MOZGIII in #732
• Add error-fmt crate by @MOZGIII in #754
• Inline the sources in the library error messages by @MOZGIII in #733
• Make out-of-range worker indices a panic in the process pool by @MOZGIII in #735
• Drop the unused default features of metrics-exporter-prometheus by @MOZGIII in #736
• VM driver hooks: a unifying trait, exposed at the transient bringups by @MOZGIII in #737
• Take the task monitor as an input of the tokio metrics bringup by @MOZGIII in #738
• Drop the unused googleapis-common-protos dependency by @MOZGIII in #739
• Run the coverage suite once for both reports by @MOZGIII in #740
• Document the observability store as one schema by @MOZGIII in #741
• Run the lossy batcher's flushers as tasks by @MOZGIII in #742
• Derive the action runtime metadata decode errors with thiserror by @MOZGIII in #743
• Log each shutdown request as one line in start-workers by @MOZGIII in #744
• Stop the durable integration run at the first early task end by @MOZGIII in #745
• Declare exactly the tokio features the remote worker pool uses by @MOZGIII in #746
• Align the transient stream missing-result status with the sibling messages by @MOZGIII in #747
• Reflow two ragged doc comments by @MOZGIII in #748
• Correct the lossy batcher's sharing doc and swapchain closers list by @MOZGIII in #749
• Take the README wording for the max concurrent instances knob by @MOZGIII in #750
• Document the worker remote bringup crates and their start error by @MOZGIII in #751
• Render the soak harness's database wait error with its optional source by @MOZGIII in #755
• Skip the lock renewal heartbeats a slow renewal held up by @MOZGIII in #756
• Declare exactly the tokio features the worker process uses by @MOZGIII in #757
• Sweep the runloop term out of the guides, scope the instances cap per process and name the HTTP variables by @MOZGIII in #758
• Truncate a timedelta timeout once, after adding up its keywords by @MOZGIII in #762
• Document the retry policy as the runtime enforces it by @MOZGIII in #761
• Delete the example app's orphan uv.lock by @MOZGIII in #759
• Derive IdDecodeError with thiserror by @MOZGIII in #760
• Log per-workflow, per-schedule and per-worker events at debug level by @MOZGIII in #763
• Update sqlx to 0.9 and switch to a git rev by @MOZGIII in #767
• Build and embed a hello-world SPA in the HTTP server by @piercefreeman in #670
• Redesign the observability webapp and wire it to the API by @piercefreeman in #704
• Read the PyPI readme from the shared README at build time by @MOZGIII in #777

Full Changelog: v0.30.dev8...v0.30.dev14