Skip to content

[review-stack 10/10] assets-tests - #16218

Open
synap5e wants to merge 2 commits into
synap5e/rs/asset-record-content-split/9-coderabbit-fixesfrom
synap5e/rs/asset-record-content-split/10-assets-tests
Open

[review-stack 10/10] assets-tests#16218
synap5e wants to merge 2 commits into
synap5e/rs/asset-record-content-split/9-coderabbit-fixesfrom
synap5e/rs/asset-record-content-split/10-assets-tests

Conversation

@synap5e

@synap5e synap5e commented Sep 10, 2026

Copy link
Copy Markdown
Contributor

Two test gaps closed, both raised in the 2026-09-08 review: the execution battery never ran with assets enabled, and list-shaped node outputs had no unit coverage.

No production changes; this layer is entirely tests.

Assets-enabled battery coverage. The execution suite boots a server per cache mode and drives real prompts over HTTP — but never with --enable-assets, so the whole asset path had no end-to-end coverage. A fourth server param now runs the same 52 tests with assets on, against an isolated database. At teardown it asserts the asset system is actually healthy: the scan settles without errors, the database and the output directory agree in both directions, and no exception-path log line was emitted.

The reconcile builds its disk set from the scanner's own enumeration and admission filters rather than a fresh walk — a looser one reports every partial download as a missing registration, a stricter one hides real misses. I checked both halves of the teardown check can fail: planting a file the API doesn't know about trips the reconcile, and an injected error line trips the log scan.

List-shaped output coverage. A node that executes N times from a list input emits N declared entries in one call. Four cases pinned: N entries register N assets under one job; mixed output/temp entries resolve against their own roots; repeated executions accumulate distinct paths without superseding each other; an empty entry list is a no-op. Each test was checked against a broken version of what it covers. The mixed-type one needs its own fixture with separate roots — the existing one points every type at the same directory, where broken per-type resolution still passes.

Suite: 1355 passed, 1 skipped. Ruff clean. 2 files, +448/−1.

+/- path
+359/-1 tests/execution/test_execution.py
+89/-0 tests-unit/execution_test/test_enrich_output.py

Stacked on #16081. Draft until the layers below land.

synap5e and others added 2 commits September 8, 2026 21:36
@synap5e
synap5e marked this pull request as ready for review September 10, 2026 04:23
@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review Completed 2026-09-10T04:28:23.163355Z 8f8ace0 Draft marked ready
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant