Update dependency prom_ex to v1.12.0 - #77
Conversation
|
🐳 Image for commit b920435 published: |
Superseded by a newer automated review for this pull request.
There was a problem hiding this comment.
AI Automated Review
Full PR review.
Analysis engine: MiniMax-M3@https://litellm.jory.dev/v1 (anthropic) — escalated (fast_low_confidence)
Recommendation
Approve. This is a clean Renovate-generated lockfile-only dependency bump (prom_ex 1.11.0 → 1.12.0, plus the matching peep transitive bump 3.5.0 → 4.4.0 that prom_ex now declares via ~> 3.0 or ~> 4.0). CI has already run the full quality gate (Lint and Test, plus the two image builds) green against this commit, and the upstream release notes contain no breaking changes affecting Pinchflat's usage.
Change-by-change findings
mix.lock — prom_ex 1.11.0 → 1.12.0
Version-pinned host/peer platform check: This is a Hex library bump, not a container-image, OS, or peer-binary bump. mix.lock does not couple prom_ex to a host OS image, an installer, or a chart-targeted operator version, so the platform-matrix step the prompt otherwise mandates does not apply. The only coupled surface here is the prom_ex ⇄ peep ⇄ finch ⇄ telemetry* chain inside Hex itself, which is governed by the new peep: "~> 3.0 or ~> 4.0" range (relaxed, not tightened) plus the unchanged finch: "~> 0.18", telemetry: ">= 1.0.0", telemetry_metrics: "~> 1.0", telemetry_metrics_prometheus_core: "~> 1.2", telemetry_poller: "~> 1.1", jason: "~> 1.4", octo_fetch: "~> 0.4" constraints — all satisfied by the versions currently locked.
Breaking-change scan against the upstream 1.11.0 → 1.12.0 diff:
Fix Oban plugin initial state hydration— bugfix only;include_zeros_for_missing_queue_states/1now takes(query_result, config)and readsconfig.plugins/config.queuesinstead of callingOban.config(). Pinchflat does not instantiate multiple Oban instances and does not subclass or wrap the Oban plugin, so the per-instance fix is irrelevant to consumers and the API change is internal to the plugin.Allow additional (custom) tags in HTTP events— additive config option on the Phoenix plugin; no existing option renamed/removed.Support Peep v4.0+— the PR message explicitly says "The APIs PromEx use didn't change at all." Pure dep-range relaxation; nothing to adapt in Pinchflat, and the lockedpeep4.4.0 satisfies the new range.Use correct legend for Atom Table Size panel in BEAM Grafana dashboard— dashboard label fix only.Keep options order in Phoenix plugin— cosmetic/internal ordering inPromEx.Plugins.Phoenix; Pinchflat does not pass an options list to that plugin, so no configuration drift risk.
The Grafana client diff swaps a Mint.TransportError match for a Finch.TransportError match — Finch-internal, not a consumer-facing API. The peep 3.5.0 → 4.4.0 lockfile bump adds a telemetry dep and otherwise keeps telemetry_metrics ~> 1.0, nimble_options ~> 1.1, and the existing optional plug ~> 1.16 — all compatible with Pinchflat's locked versions. No public API surfaces Pinchflat touches are affected.
What Pinchflat actually uses from prom_ex
The repo impact scan found no grep hits for 1.11.0, the changelog anchor, or akoutmos/prom_ex anywhere in lib/, test/, config/, or docker/. prom_ex is wired in for observability dashboards; the upgrade does not touch config, schema, workers, or any endpoint. Repository-history grep also shows no in-repo PR or commit message referencing the bumped PR numbers (259, 270, 272, 278, 286), so there are no local call sites that could be relying on changed behaviour.
Required checks
- Check for breaking API changes in updated dependencies. Verified. Walked the upstream 1.11.0 → 1.12.0 diff and changelog; every change is either an internal bugfix (Oban plugin hydration, Grafana transport error match), an additive config option (custom HTTP tags), a dashboard label fix, a cosmetic ordering fix, or a transitive range relaxation (
peep ~> 3.0 or ~> 4.0). The maintainer's note on the Peep bump explicitly states no API changes. Pinchflat does not customise the affected surfaces. No breaking changes identified. - Run full test suite after upgrade. Verified via authoritative CI evidence. The CI status API reports
Lint and Test,Build RC Image, andBuild PR Imageall success against this PR's head commit9d32507. Per the standards file,Lint and Testrunsmix check --no-fix --no-retryinside the pinnedghcr.io/misospace/pinchflat-ci-baseimage withEX_CHECK=1, which exercisesmix deps.audit,mix sobelow,mix format --check-formatted(implicit),mix credo,mix compile --warnings-as-errors,yarn run lint:check, andmix test(with the required./tooling/fetch-sqlean.shfirst). The two image builds also exercise the production compile path. The corpus explicitly notes "All four are green today (1208 tests, 0 failures)" as the standing baseline. No flakes attributable to the upgrade.
Standards Compliance
The standards file's mix check gate is the protected-branch gate, and CI confirms it passed. No documentation update is required: AGENTS.md, CODEBASE.md, and DEVELOPMENT.md describe behaviour and config surfaces; none of those change here — prom_ex configuration, dashboards, and metrics are not documented as user-facing behaviour in any of those files. A minor-version library bump with no behaviour, config, endpoint, or capability change does not trigger the "PR must update that doc" rule. tooling/.check.exs and tooling/.credo.exs are not in scope of the root formatter per the standards file; no changes touch them. No new compiler warnings expected (CI's --warnings-as-errors flag is green).
Unknowns or Needs Verification
None blocking. Minor caveats the reviewer should still be comfortable with, but which do not change the verdict:
- The upstream release date stamps 2026-06-19, and several commits in the compare view also bear 2026 dates — the repo's
CHANGELOG.md1.12.0apprise line is also dated far in the future relative to the cutoff implied by the standards file's references. This is a date-skew artifact of the upstream metadata, not a Pinchflat issue. - If Pinchflat ever subclasses
PromEx.Plugins.ObanorPromEx.Plugins.Phoenix, theinclude_zeros_for_missing_queue_states/2arity change and the options-order reshuffle in the Phoenix plugin would matter. The repo impact scan found no such subclassing, so this is a forward-looking note only.
Sources
- Upstream 1.12.0 compare: upstream akoutmos/prom_ex compare 1.11.0...1.12.0
- Upstream 1.12.0 release: https://github.com/akoutmos/prom_ex/releases/tag/1.12.0
- CI check results (authoritative):
Lint and Test,Build RC Image,Build PR Imageall success on commit 9d32507
|
🐳 Image for commit 9d32507 published: |
This PR contains the following updates:
1.11.0→1.12.0Release Notes
akoutmos/prom_ex (prom_ex)
v1.12.0Compare Source
Configuration
📅 Schedule: (UTC)
* * 1,15 * *)🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about this update again.
This PR has been generated by Mend Renovate CLI.