feat(memory): replace LanceDB with tantivy-jieba full-text search#160

Merged

Bahtya merged 21 commits intomainfrom

feat/tantivy-memory-main

Apr 24, 2026

Commits on Apr 23, 2026

milestone: complete comparative analysis of KA vs Hermes memory systems
Bahtya
committed
feat: replace LanceDB with tantivy-jieba full-text search
Bahtya
committed
fix: update tantivy_store.rs for tantivy 0.26 API and remove dead code
Bahtya
committed
fix: JiebaTokenizer constructor, collector reference, and formatting
Bahtya
committed
fix: remove embedding from gateway/heartbeat, use TantivyStore, remove dead schema field
Bahtya
committed
fix: format execute_learning_actions call in test
Bahtya
committed
fix: format execute_learning_actions on single line
Bahtya
committed
fix: add futures to dev-dependencies for concurrent test
Bahtya
committed
fix: add LowerCaser filter to jieba tokenizer for case-insensitive search
Bahtya
committed
fix: use explicit lowercasing for case-insensitive search with jieba
Bahtya
committed
fix: separate stored content from search index to preserve original case
Bahtya
committed
style: format QueryParser::for_index on single line
Bahtya
committed
style: fix rustfmt formatting for content_search fields
Bahtya
committed
style: match rustfmt chain formatting for content_search_field
Bahtya
committed
fix: keep tempdir alive in heartbeat memory tests
Bahtya
committed
fix: skip capacity check for upserts (overwrite existing entries)
Bahtya
committed
refactor: use TextAnalyzer+LowerCaser for case-insensitive search
Bahtya
committed
style: fix rustfmt formatting for chain calls
Bahtya
committed
refactor: adopt CC-Adv review feedback — pure read recall + single field
Bahtya
committed
style: single-line QueryParser call per rustfmt
Bahtya
committed
fix: update test to match pure-read recall behavior
Bahtya
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(memory): replace LanceDB with tantivy-jieba full-text search#160

feat(memory): replace LanceDB with tantivy-jieba full-text search#160
Bahtya merged 21 commits intomainfrom
feat/tantivy-memory-main

Commits on Apr 23, 2026

milestone: complete comparative analysis of KA vs Hermes memory systems

feat: replace LanceDB with tantivy-jieba full-text search

fix: update tantivy_store.rs for tantivy 0.26 API and remove dead code

fix: JiebaTokenizer constructor, collector reference, and formatting

fix: remove embedding from gateway/heartbeat, use TantivyStore, remove dead schema field

fix: format execute_learning_actions call in test

fix: format execute_learning_actions on single line

fix: add futures to dev-dependencies for concurrent test

fix: add LowerCaser filter to jieba tokenizer for case-insensitive search

fix: use explicit lowercasing for case-insensitive search with jieba

fix: separate stored content from search index to preserve original case

style: format QueryParser::for_index on single line

style: fix rustfmt formatting for content_search fields

style: match rustfmt chain formatting for content_search_field

fix: keep tempdir alive in heartbeat memory tests

fix: skip capacity check for upserts (overwrite existing entries)

refactor: use TextAnalyzer+LowerCaser for case-insensitive search

style: fix rustfmt formatting for chain calls

refactor: adopt CC-Adv review feedback — pure read recall + single field

style: single-line QueryParser call per rustfmt

fix: update test to match pure-read recall behavior