fix(results): snapshot validated application authority

[seonghobae]

ACTIVE Draft — bounded result-application authority + semantic core

Protected main@b84f0c94154043a3473939c01bb6471de5a129ae remains unchanged. This lane is an explicit non-destructive child of coverage-root #233@5951b7a4d779903b8924abaef2a387cae50b7f54. Current exact head is 5d9407a234794ed02444b0f143bba80af7d09045.

Bounded authority repair

The branch captures caller-owned result/checkpoint authority before store/callback hooks, reconstructs exact package-owned checkpoint snapshots, recursively detaches exact JSON primitives under finite structural/text budgets, rejects behavior-bearing/malformed/cyclic/non-finite authority, preserves transaction-local cursor revocation and checkpoint conflict semantics, and does not claim distributed exactly-once delivery.

The resource-boundary repairs are test-first:

• predecessor RED ef233fe94114ab216adae444cf7503d204168979 proved multibyte text could exceed the intended DEFAULT_MAX_JSONL_LINE_BYTES budget when Python code points were counted rather than UTF-8 bytes; GREEN f2c94afd4a068fb27db33cb560399c4cf9e0cbce performs bounded chunked UTF-8 byte accounting;
• RED 298ffc13d28ef6f9ef2a281a95ce3e6f98a70b51 proved exact integer JSON values consumed no text budget ({"n":12345} survived a four-byte snapshot budget);
• RED 1b1db1ae6ba4923556bac40c512cfd6f8a6b504b additionally requires Python's oversized integer-to-string conversion limit to fail through the same redacted package validation boundary;
• current GREEN 5d9407a234794ed02444b0f143bba80af7d09045 counts decimal integer representation against the finite text budget and converts ValueError from oversized integer string conversion into bounded ValidationError evidence.

This prevents caller/post-construction-mutated numeric values from bypassing the same finite decoded-record resource boundary enforced for text. The error remains content-redacted (item.record, <redacted>).

Semantic naming / compatibility

The useful source/test delta from later naming Draft #324 remains folded into this earlier Result Application authority owner. Package-owned implementation uses semantic multiword names such as transaction_cursor, checkpointed_record, record_effect, previous_checkpoint, result_checkpoint, field_name, and validation_reason, with _SemanticResultApplicationOutcome(record_applied, result_checkpoint) behind the released compatibility surface.

Public compatibility remains test-pinned for ResultApplicationError(phase=...), ResultApplicationOutcome(applied, checkpoint), and apply_checkpointed_result_in_transaction(cursor, checkpoint_store, consumer_name, item, apply_record).

Descendant ancestry

#324 is documentation-only. After this parent advanced, it was non-force restacked to exact head 69520a8a6944ad12b59068739602f674e7f79cab. Fresh comparison from this exact parent reports behind_by=0 and merge base exactly 5d9407a...; the child diff remains only ARCHITECTURE.md, CHANGELOG.md, docs/doctoring/result-application-semantic-identifiers.md, and docs/product-technical-gap-baseline.md. No predecessor checks/reviews transfer.

Exact-current evidence

Exact-head CI 33659906131 and Release Acceptance 33659906071 are still non-passing. CI materialized all five expected jobs on 5d9407a..., but each is queued before checkout with runner_id=0, no runner/group identity and steps=[]: container/PostgreSQL 100347641140, coverage/docstrings/lint/package 100347641486, Python 3.12 100347641563, Python 3.10 100347641601, and Python 3.14 100347641657.

This is runner-acquisition/control-plane evidence, not a source failure and not hosted GREEN. Exact evidence was advanced to existing central owner issue .github#712; no leaf churn or gate weakening is justified merely to retrigger admission.

Fresh formal reviews and inline review threads are empty on this exact head. Absence of review is not approval.

Stack and governance

#233 remains the exact predecessor and separately governed. Keep this PR Draft until #233 integrates or is genuinely superseded and this head is reconciled non-destructively to the then-current protected main. Before Ready/merge, regenerate actual-checkout exact-head evidence for supported Python including 3.14, exact 100% owned production statement/branch coverage, 100% public docstrings, PostgreSQL/container/package/reproducibility/security/SAST/release acceptance, zero valid findings/threads, and every then-live governance requirement.

Queued, pending, cancelled, skipped-required, absent, neutral, stale, predecessor, status-only, synthetic, author-only, rate-limited, infrastructure-failed, dismissed, or conclusion-null evidence is non-passing. No force push, destructive rebase, self-approval, gate weakening, or central workaround copy.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}