fix: address 6 confirmed review issues on PR #514#520
Conversation
…ompat The multi-language eval refactor required source files in detect() methods, breaking discovery which only needs marker files (package.json, Cargo.toml, etc.) to identify project language. Phantom detection prevention is already handled by each run_*() method returning None when no source files exist. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…or paths Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- EvalFragment: add coverage_pct field to fix passed semantic overload - NodeEvaluator.name: return typescript unconditionally (fix detection regression + statefulness) - Rust: revert cargo test --workspace to cargo test - Go/Rust/Node: stub new eval dimensions to None (behavioral parity with main) - Tests: update characterization tests to reflect all fixes Closes #519 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #520 +/- ##
==========================================
+ Coverage 86.77% 87.92% +1.14%
==========================================
Files 64 70 +6
Lines 10027 10120 +93
==========================================
+ Hits 8701 8898 +197
+ Misses 1326 1222 -104 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Filter out EvalFragment entries where coverage_pct is None before summing, preventing a TypeError crash if a future evaluator returns a coverage fragment without setting coverage_pct. Also handle the empty-list edge case to avoid ZeroDivisionError. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…priority Swap node and rust registration so detection priority is python → node/typescript → rust → go, matching the original _detect_language behavior. Without this, projects with both Cargo.toml and package.json would incorrectly detect as 'rust'. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
✅ Factory Review: KEEPVerdict: KEEP Experiment: #20 Score Comparison
Guard Checks
Posted by Factory CEO |
✅ Factory Review: KEEPVerdict: KEEP Experiment: #20 Score Comparison
Guard Checks
Posted by Factory CEO |
|
@osilkin98 it made a new PR. I filed a bug report about this. But please take a look |
|
@colehurwitz Thank you for addressing the changes, sorry it's been such a hassle. Let's merge this and fix the other bug. |
Closes #519. Addresses review feedback from @osilkin98 on PR #514.
Summary
Fixes all 6 issues confirmed in the code review:
This PR is a behavior-preserving architecture refactor. New eval dimensions for Go/Rust/Node will be enabled in a follow-up PR with explicit disclosure.
Refs: PR #514, #513