4 Commits

Author SHA1 Message Date
Michael A. Kuykendall
0f0b95a8d4 fix: resolve all test failures from Issue #132 merge conflict
Issue #132 added stop field to ChatCompletionRequest and stop_tokens to GenOptions.
Test code wasn't updated, causing CI PPT Contract Tests to fail.

Changes:
1. Added stop: None to 13 ChatCompletionRequest test initializers:
   - src/openai_compat.rs (7 locations)
   - tests/openai_api_real_tests.rs (6 locations)
   - tests/api_error_handling_test.rs (1 location)

2. Added stop_tokens: Vec::new() to 2 GenOptions test initializers:
   - src/engine/universal.rs (1 location)
   - src/engine/huggingface.rs (1 location)

3. Fixed release.yml gate numbering inconsistency:
   - Changed GATE 6/8 -> GATE 6/7
   - Changed GATE 7/8 -> GATE 7/7
   (There are only 7 gates total, not 8)

4. Fixed workflow.rs to validate output steps exist:
   - Now checks that all requested outputs reference actual steps
   - Sets success=false when output steps are missing
   - Provides error message listing missing steps

All 445 tests now passing (3 ignored).
Release gate validation: PASS (10/10)

Signed-off-by: Michael A. Kuykendall <michaelallenkuykendall@gmail.com>
2025-10-22 16:39:44 -05:00
Michael A. Kuykendall
6842f05d0c fix: resolve test warning for unused response variable 2025-10-09 12:42:13 -05:00
Michael A. Kuykendall
8273191b34 refactor: Remove unused specs and dead code, fix clippy errors
- Removed specs: 001-model-preloading, 002-response-caching, 004-request-routing
- Removed dead code: ModelCache, RouteManager, test_cache_performance binary
- Fixed clippy redundant closure warnings in llama.rs
- Fixed missing Default impl for ObservabilityManager
- Removed unused fields from OptimizationState
- Cleaned up unused imports in cache/mod.rs
- Removed routing module references from lib.rs and main.rs

All clippy checks passing, code formatted, ready for release.
2025-10-03 15:16:37 -05:00
Mike Kuykendall
a9ffd51a79 fix: Improve error handling for missing models in OpenAI API compatibility
Better error response format for Issue #65: Windows 404 Error

- Enhanced error handling in src/openai_compat.rs to return JSON error responses
- Added comprehensive regression test for OpenAI-compatible error format
- Error responses now include available models list for better user experience
- Maintains OpenAI API compatibility with proper error structure

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-09-22 16:04:52 -04:00