3.6.5

• fix table creation in put() and put_batch()

3.6.3

Graph Sharing via Relay:

• Share your local graphs instantly with g.serve(share=True)
• Get a public URL (e.g., https://abc123.s.cogdb.io/) to access your graphs from anywhere
• No port forwarding or firewall configuration required, works behind NAT
• Remote Graph Connections
• Connect to shared graphs using Graph.connect(url)

from cog.torque import Graph
g = Graph("mydb")
g.put("alice", "knows", "bob")
g.serve(share=True)

3.6.1

Adding more tests.

3.6.0

• Enhanced Tagging Functionality: The tag() method now supports applying multiple tags simultaneously by accepting a list of tag names, while maintaining backward compatibility for single string tags.
• New Vertex Ordering Method: A new order() method has been introduced to sort vertices by their ID in either ascending or descending order, providing more control over traversal results.
• Sorting Direction Constants: New constants, asc and desc, have been added to cog.torque to clearly specify the desired sorting direction for the order() method.
• Comprehensive Unit Tests: Extensive unit tests have been added to validate the new multi-tagging capabilities and the vertex ordering functionality, covering various usage scenarios and edge cases.

3.5.1

Fix edge update logic and index deletion edge cases

• Refined update_edge to preserve unrelated edges during updates
• Improved index deletion for hash collision chain scenarios
• Added tests for edge operations

3.5.0

🚀 CogDB v3.5.0 – Graph Server Support

This release introduces HTTP server capabilities for CogDB, enabling remote access to your graph databases over the network.

• Serve any graph over HTTP with a single method call: graph.serve()
• Multi-graph support: serve multiple graphs on the same port with path-based routing
• Thread-per-request architecture using stdlib http.server (zero additional dependencies)
• Simple UI for graph endpoint

3.4.4

What's Changed

• feat: add bfs() and dfs() graph traversal methods by @arun1729 in #62

3.4.3

Adding pure Python cosine distance fallback.

3.4.2

v3.4.2 Release Notes

Bug Fixes

• Scanner collision handling: Index.scanner() now correctly iterates through all records in hash collision buckets by following key_link chains.
• Embedding storage: Removed redundant pre-hashing that caused data loss at scale (1.2M+ embeddings).
• k_nearest semantics: Properly distinguishes between “no traversal” vs. “empty traversal result”.

Performance

• Full GloVe vocabulary (1.2M words) now loads correctly.

Tests

• Added collision test suite with small INDEX_CAPACITY.

3.4.1

v3.4.1: relax xxhash version constraint to fix dependency conflicts