Skip to main content
Glama
IRIS_VECTOR_RAG_0.5.4_SUMMARY.md1.83 kB
# iris-vector-rag 0.5.4 Testing Summary **Date**: December 14, 2025 **Version**: 0.5.4 (local unreleased build) **Installation**: `/Users/tdyar/ws/iris-vector-rag-private/dist/iris_vector_rag-0.5.4-py3-none-any.whl` ## Quick Results **Tests Passed**: 4/6 ✅ **Status**: CONNECTION BUG FIXED, but dimension bug returned ## Version Comparison | Version | Tests Passed | Connection | Dimension | Status | |---------|-------------|------------|-----------|--------| | v0.5.2 | 4/6 | ✅ (workaround) | ❌ | Original baseline | | v0.5.3 | 3/6 | ❌ BROKEN | ✅ FIXED | Critical regression | | v0.5.4 | 4/6 | ✅ FIXED! | ❌ RETURNED | Back to baseline | ## What Changed ### ✅ Connection Bug FIXED - v0.5.3 used `iris.connect(host, port, ...)` - positional params (WRONG) - v0.5.4 uses `iris.connect(hostname=host, port=port, ...)` - named params (CORRECT) - **Result**: All connection-dependent tests now pass ### ❌ Dimension Bug RETURNED - v0.5.3 CloudConfiguration API **worked** (returned 1024) - v0.5.4 CloudConfiguration API code **present but broken** (returns 384) - **Cause**: Unknown - needs investigation ## Recommendation **Use v0.5.4**: Connection is critical - v0.5.4 is better than v0.5.3 despite dimension regression. **Workaround for dimensions**: Continue using our `IRISVectorDBClient` implementation which handles dimensions correctly. ## Files Created 1. `IRIS_VECTOR_RAG_0.5.4_FINDINGS.md` - Comprehensive analysis with code examples 2. `PROGRESS.md` - Updated with v0.5.4 testing section 3. This summary file ## Next Actions - Share findings with iris-vector-rag team - Request investigation of CloudConfiguration API regression - Monitor for v0.5.5 with dimension fix --- **Overall Assessment**: v0.5.4 is a step forward (connection restored) but still needs work (dimension regression).

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/isc-tdyar/medical-graphrag-assistant'

If you have feedback or need assistance with the MCP directory API, please join our Discord server