dan/memory
1
0
Fork 0
mirror of https://github.com/mruwnik/memory.git synced 2026-01-02 09:12:58 +01:00
Code Issues 1 Packages Projects Releases Wiki Activity

Update bug classifications in INVESTIGATION.md

Browse Source 
- BUG-023: Mark as acceptable design (SHA256 for exact dedup)
- BUG-026: Mark as acceptable design (BM25/embedding for recall, LLM for ranking)

These are architectural decisions, not bugs - semantic dedup and hybrid scoring would be feature enhancements.

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

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
mruwnik 2025-12-19 21:46:52 +00:00
parent be3963b02f
commit 92dad5b9fd
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
INVESTIGATION.md
View File

@ -317,12 +317,12 @@ Based on git history analysis, the following bugs have been FIXED:
### Content Processing ### Content Processing
- BUG-021: ✅ No chunk validation after break_chunk - FIXED (yield_spans guarantees max_tokens) - BUG-021: ✅ No chunk validation after break_chunk - FIXED (yield_spans guarantees max_tokens)
- BUG-022: Low priority - extract_ebook creates single chunk, BUT sync_book task properly creates BookSection chunks - BUG-022: Low priority - extract_ebook creates single chunk, BUT sync_book task properly creates BookSection chunks
- BUG-023: SHA256-only deduplication misses semantic duplicates (`source_item.py:51-91`) - BUG-023: Acceptable design - SHA256 is for exact deduplication (semantic dedup would be a feature enhancement)
- BUG-024: Email hash inconsistency with markdown conversion (`email.py:171-185`) - BUG-024: Email hash inconsistency with markdown conversion (`email.py:171-185`)
- BUG-025: Acceptable - 4 chars/token is common approximation (accurate tokenization requires model-specific tokenizers) - BUG-025: Acceptable - 4 chars/token is common approximation (accurate tokenization requires model-specific tokenizers)
### Search System ### Search System
- BUG-026: BM25 scores calculated then discarded (`bm25.py:66-70`) - BUG-026: Acceptable design - BM25/embedding scores used for candidate selection, LLM scorer provides final ranking (hybrid scoring would be an enhancement, not a bug fix)
- BUG-027: N/A LLM score fallback - actually reasonable (0.0 means chunk not prioritized when scoring fails) - BUG-027: N/A LLM score fallback - actually reasonable (0.0 means chunk not prioritized when scoring fails)
- BUG-028: ✅ Missing filter validation - FIXED (unknown filter keys now logged and ignored instead of passed through) - BUG-028: ✅ Missing filter validation - FIXED (unknown filter keys now logged and ignored instead of passed through)
- BUG-029: N/A Hardcoded min_score thresholds - intentional (0.25 text, 0.4 multimodal due to different score distributions) - BUG-029: N/A Hardcoded min_score thresholds - intentional (0.25 text, 0.4 multimodal due to different score distributions)