alkblobs/tasks/alkcall-010-bump.md at v0.1.0

0 Fork 0

Files

T

glm-5.3-flash a57558899b chore(deps): alkcall 0.9 -> 0.10 — the gateway ownership-seam bump (wave 13, review-002 step 1)

- Cargo.toml: alkcall pin 0.9 -> 0.10 (optional, ops feature)
- No source changes: alkblobs rides the native dispatch wire with
  ownership providers wired, so the loud FORBIDDEN default and the new
  GatewayDispatch builder methods (with_ownership_provider,
  with_static_ownership_fallback) are outside this crate's path -
  the 0.10 survey's "no impact" conclusion held
- Gates: cargo test (default) green; cargo test --all-features green
  against the live pg harness (460 passed); clippy -D warnings clean
  on both feature variants; cargo fmt --check clean; local Cargo.lock
  relocked to alkcall v0.10.0 (lock is gitignored, not tracked)
- task: alkcall-010-bump flipped to completed

2026-10-17 18:11:03 +00:00

3.1 KiB

Raw Permalink Blame History

id, name, status, depends_on, scope, risk, impact, level, tags

name

status

depends_on

scope

risk

impact

level

Description

alkcall published 0.10.0 (the gateway ownership seam — alkcall review 012, downstream evidence from alkgit's POC-L): additive API (GatewayDispatch::with_ownership_provider, with_static_ownership_fallback) plus one behavior change — the loud ownership default (FORBIDDEN, pre-dispatch, for resource_type-gated ops on a spine with no ownership provider wired and no static fallback declared).

alkblobs is a producer on the native wire — the ops module's ACL mapping rides the native dispatch path with providers wired, not a gateway door — and alkcall's own 0.10 survey found no impact for this crate. So the expected change is the manifest pin alone:

Cargo.toml: alkcall = { version = "0.10", optional = true }

Then re-run the gates and confirm the ops surface compiles and passes clean: the ops e2e flows, the ACL matrix (closed-by-default), the token renewal forms, the put/fetch pipes. No code change is expected; if the loud default or the new builder methods force any adaptation, it lands here with the reason recorded in Notes (and the review-002 record gains a line — the survey's "no impact" conclusion would have been wrong).

Acceptance Criteria

Cargo.toml pins alkcall = "0.10"
The ops-bearing test runs green unmodified (e2e flows, ACL matrix, token forms) — or the needed adaptation landed with its reason recorded
Gates green: cargo test (default), cargo test --all-features (server-less; the pg columns skip cleanly), clippy --all-targets -- -D warnings across the feature variants, cargo fmt --check
Cargo.lock updated consistently

References

docs/reviews/002-first-consumer-review.md (the round's context; the bump is the endorsed sequencing's step 1)
/workspace/@alkdev/alkcall/CHANGELOG.md ([0.10.0] — the ownership seam, the loud default, the survey's consumer-impact findings)
Cargo.toml (the ops feature's alkcall pin)

Notes

Ran against the live pg harness (ALKBLOBS_PG_* set to the dockerized pglo-poc server on :15432) for the --all-features gate, so the pg/pg-lo/fleet columns ran for real rather than skipping.
No code adaptation was needed — the loud ownership default and the two new builder methods touch the gateway dispatch spine only, and alkblobs rides the native wire with providers wired. The survey's "no impact" conclusion held; review-002 needs no further line past this bump.

Summary

The manifest-only bump landed as expected: alkcall = "0.10" in Cargo.toml, Cargo.lock relocked to alkcall v0.10.0, zero source changes to the ops module. Verified with the full gate block: default cargo test green; cargo test --all-features green against the live pg harness (460 passed across the suites, ~107s/96s on the pg-heavy columns); cargo clippy -D warnings clean on both the default and --all-features target sets; cargo fmt --check clean.

3.1 KiB Raw Permalink Blame History