Skip to content

Fix audit regressions and verify installed artifacts - #70

Merged
aurokin merged 3 commits into
mainfrom
fix/audit-recovery-and-verification
Oct 8, 2026
Merged

aurokin merged 3 commits into
mainfrom
fix/audit-recovery-and-verification

Conversation

@aurokin

@aurokin aurokin commented Oct 7, 2026 •

Copy link
Copy Markdown
Owner

Partial guide refreshes could overwrite bundles from completed comparisons, SimC comparisons could overwrite their inputs, and sampled analytics could combine different pulls or seasons. This patch fixes all nine audit defects, preserves earlier guide export generations, and moves Warcraft Logs discovery/readiness and Wowhead guide retrieval/export out of CLI modules. Composed guide exports now use provider functions directly.

  • Correct effective SimC APL replacement, inline and literal-append semantics; protect all generated profile/report paths before writes; preserve Raidbots first-actor talent overrides.
  • Apply canonical/localized WCL realm fallback to reports; keep different keystone levels and contradictory rosters separate; pin Raider.IO sampling seasons.
  • Preserve Wowhead graph edges at node limits and complete comment citation sources; resolve provider cache roots at settings load; isolate missing-auth E2E tests from repository dotenv files.
  • Verify the installed root wheel outside the checkout in CI and releases; fix pip deployment from other directories; update behavior docs and Unreleased notes.

Validation:

  • Local make check with the rebuilt, matching SimC binary passed: 3,599 tests and 94.78% coverage; two intentionally inapplicable HTML-provider JSON cases skipped.
  • All 16 real-binary tests passed, including four new effective-APL runtime regressions. The full real harness/validation/comparison/report journey also passes and verifies input collision rejection preserves the source APL.
  • Installed-wheel checks passed for all 12 binaries, schema output, package origins and bundled Icy Veins data. All 15 GitHub checks passed on the final commit.
  • Diffwarden's closing unscoped review found no issues. Its initial APL separator/fixture findings are fixed and covered by runtime tests.
  • Full live E2E finished with 563 passed, 77 failed and 2 explicit exclusions. Six failures were fixture assumptions and are fixed: Lorrgs 28/28, Raider.IO 63/63, SimC workflow 1/1 and guide refresh/handoff 2/2 pass on rerun. The remaining 71 failures are WCL or dependent wrapper journeys blocked by rejected configured authentication (auth_failed, exit3); those live reads remain unverified. Redis and public Raidbots-report journeys are excluded because their required inputs are unconfigured.

aurokin and others added 2 commits October 7, 2026 17:41
Protect guide generations and SimC inputs, correct effective APL and sampled cohort semantics, extract provider services, and verify packaged and live behavior.

Co-Authored-By: OpenAI Codex <noreply@openai.com>
Select a loaded Lorrgs report for report journeys, assert inclusive Raider.IO filters against observed values, and keep SimC workflow sources separate from generated outputs while verifying collision rejection.

Co-Authored-By: OpenAI Codex <noreply@openai.com>
@aurokin
aurokin marked this pull request as ready for review October 7, 2026 23:57
Exercise copied-root reuse and immutable generations without rewriting the module fixture used by later SimC handoff journeys.

Co-Authored-By: OpenAI Codex <noreply@openai.com>
@aurokin
aurokin merged commit fcd691d into main Oct 8, 2026
15 checks passed
@aurokin
aurokin deleted the fix/audit-recovery-and-verification branch October 8, 2026 00:03
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