zeekay
1b367e2b62
The tree was a debranded snapshot of upstream taken around 2026-02-20, and upstream has moved about five months since. This replaces it with upstream master, mechanically re-derived: bin/debrand does the rename, and the pipeline and squashed migrations are restored from the shipped tree. Not a merge. The previous import replaced the tree wholesale in a single commit, so there is no shared content history to merge against. Reproducing the rename on the new tree is what actually converges. What survives: .hanzo/workflows, the Dockerfiles and bin/docker* entrypoints, compose files, LLM.md, and insights/migrations. The squashed initial migration is load-bearing — the live database records it by name, so upstream's ~918 individual migrations cannot come back. Nine products are dropped as unreferenced. Twenty-one more that we do not run are kept anyway because the core app still imports them; unpicking those is a refactor and does not belong in a rebase. This base does not build yet, and the reasons are not in the rename: upstream now wants Python 3.13.13 against our 3.12.12 and Django 5.2 against our 4.2, and the dependency manifests still name upstream packages that we republish under our own names. Those are decisions, not renames.
1.6 KiB
paths
| paths | ||
|---|---|---|
|
Editing a @workflow.defn body (adding/removing/reordering execute_activity, child-workflow
starts, or timers) breaks in-flight executions with "Non Deterministic Error" — on deploy every
running execution replays its recorded history against the new code, and any new unconditional
command at a point an execution already passed fails the replay. Activity implementations and
activity input dataclasses are safe to edit; workflow command sequences are not.
When adding a new activity or child-workflow start to an existing workflow, gate it one of two ways:
if workflow.patched("my-change-2026-07"): ...— the default. Old histories skip the block, new executions record a marker and run it. Precedent:insights/temporal/ai_observability/run_evaluation.py. Once no pre-patch executions remain (days, for short-lived workflows), optionally swap toworkflow.deprecate_patch(...)+ unconditional code.- Gate on a new field of an existing activity's output dataclass that defaults to the skip
value (e.g.
enrichment_needed: bool = Falseincreate_job_model.py) — old histories decode the missing field to the default, so the new command never fires during replay. Only valid when the decision comes from recorded history and skipping is acceptable for in-flight runs.
Never gate workflow commands on values computed inside the workflow body (feature flags, settings,
clock, DB reads) — that is itself non-deterministic. Removing/reordering existing commands needs
workflow.patched() with the old path kept in the else branch.