chromadb 0.5.7

on Python PyPI

What's Changed

• [ENH] Add support for FE tolerations by @HammadB in #2560
• [ENH] Support FE replica count by @HammadB in #2561
• [ENH] fix types export for CJS consumers/NPM types badge by @codetheweb in #2563
• [ENH] CIP: Write-Ahead Log Pruning & Vacuuming by @codetheweb in #2498
• [BUG]: Pydantic 1.9+ compatibility by @tazarov in #2229
• [BUG] HNSW Rust params should have defaults by @HammadB in #2564
• [ENH] enable retries on all gRPC channels by @codetheweb in #2572
• [TST] add hypothesis collection id to logs by @sanketkedia in #2573
• [ENH] return chroma-trace-id header, include trace ID in thrown errors by @codetheweb in #2570
• [ENH] Connection pool FD leak v2 by @tazarov in #2014
• [ENH] Properly display memberlist in SysDB by @Ishiihara in #2565
• [CHORE] Add logs to memberlist by @HammadB in #2583
• [CHORE] Add more debugs to memberlist by @HammadB in #2584
• [PERF] Trace auth requests by @HammadB in #2589
• [ENH] simplify logic for when to persist index changes (re-apply with fix) by @codetheweb in #2545
• [ENH] add .clean_log() to Producers by @codetheweb in #2549
• [ENH] automatically clean WAL by @codetheweb in #2557
• [ENH] add vacuum CLI command by @codetheweb in #2519
• [BUG]: Pydantic 1.x Union Types by @tazarov in #2577
• [PERF] Add sysdb catalog traces by @HammadB in #2593
• [ENH] Add sysdb cpu/memory requests and limits to helm chart by @HammadB in #2596
• [BUG] Sysdb k8s resource typo by @HammadB in #2600
• [ENH] JS Client Refactor (Full Stack) by @AlabasterAxe in #2542
• [ENH] add tracing to gorm by @codetheweb in #2606
• [ENH] propogate nested error message by @codetheweb in #2608
• [BUG] fix nil dereference during error log by @codetheweb in #2609
• [ENH] prefetch APIs for Record segment and blockfile by @sanketkedia in #2547
• [ENH] Dispatch prefetch() as operator by @sanketkedia in #2562
• [ENH] Tag operators as IO or other + dispatcher schedules IO tasks on the main runtime by @sanketkedia in #2603
• [TST] Isolated Test Environments using testcontainers by @AlabasterAxe in #2607
• [ENH] multipart S3 file uploads by @codetheweb in #2590
• [CLN] Refactor rust codebase into many crates by @HammadB in #2618
• [ENH]: only use multipart upload if object size > part size by @codetheweb in #2622
• [CLN] Move distance, and index into their own crates by @HammadB in #2624
• [BUG]: fix nightly test workflow by @codetheweb in #2635
• [BLD]: fix JS dev release workflow by @codetheweb in #2636
• [BUG] fix multi collection log purge by @codetheweb in #2617
• [ENH]: don't allow a null collection ID on segments by @codetheweb in #2626
• [CHORE] Move pull log trace record count to structured field by @HammadB in #2642
• [PERF] Make pull logs an I/O operator so it runs on main runtime by @HammadB in #2641
• [BUG] S3 client not respecting timeout by @HammadB in #2643
• [ENH] gRPC segment requests should include the collection ID by @codetheweb in #2613
• [PERF]: bump compaction partition size to 10k by @codetheweb in #2647
• [ENH]: allow configuring replica count of compaction service by @codetheweb in #2650
• [BUG]: fix tracing for IO tasks by @codetheweb in #2652
• [ENH]: fix compactor traces, add additional info to trace by @codetheweb in #2653
• [ENH]: add record count to span when submitting embeddings by @codetheweb in #2654
• [ENH] fix embedding validation span pollution by @codetheweb in #2658
• [ENH]: lift frontend gRPC retry to application & trace by @codetheweb in #2659
• [ENH] Fetch blocks in parallel and make load_blocks take a slice instead by @HammadB in #2662
• [CLN] Only use distance avx/neon/sse if enabled by @HammadB in #2664
• [BLD]: fix nightly tests pod log upload by @codetheweb in #2670
• [ENH] Make Rust/C++ FFI error handling robust by @HammadB in #2667
• [ENH] Add timeouts to FE grpc clients by @HammadB in #2671
• [ENH]: allow using node selectors for pods instead of tolerations by @codetheweb in #2672
• [ENH] [PERF] add indices to metadata columns by @levand in #2623
• [BLD]: bump time dependency by @codetheweb in #2679
• [ENH]: skip brute force search when log is empty by @codetheweb in #2666
• [ENH]: allow configuring replicaCount and nodeSelector on log service by @codetheweb in #2678
• [ENH]: improve compaction traces by @codetheweb in #2680
• [ENH] Network Admission Control APIs by @sanketkedia in #2628
• [ENH] Construct and pass NAC by @sanketkedia in #2630
• [ENH] Block manager and HNSW provider consume Network Admission Control by @sanketkedia in #2631
• [ENH] NAC rate limits requests by @sanketkedia in #2632
• [ENH]: use follows_from() standard instead of manually adding parent attribute by @codetheweb in #2691
• [ENH]: cleanup HNSW temporary files after registering result by @codetheweb in #2646
• [PERF] Multipart s3 downloads passing through NAC by @sanketkedia in #2698
• [CLN] Reorganize delta module into seperate module and split out impls and [PERF] Refactor bf get_size to avoid nested loops by @HammadB in #2674
• [ENH] Evict other versions of hnsw index of the collection when another version is fetched by @sanketkedia in #2707
• [ENH]: add metric for total # of uncompacted log records by @codetheweb in #2696
• [ENH] Add gc for log service by @Ishiihara in #2634
• [BUG]: fix log service metrics service name by @codetheweb in #2713
• [PERF] Binary search get_prefix() calls in blockfile by @sanketkedia in #2717
• [ENH]: update JS CloudClient to accept tenant parameter by @codetheweb in #2726
• [ENH]: add config fields for max partition size & max compaction size in compactor by @codetheweb in #2732
• [PERF] Decrease compaction RAM usage and increase speed by @sanketkedia in #2729
• [PERF]: remove mutex around tokenizer by @codetheweb in #2735
• [PERF]: better locking of uncommitted tracking maps (decrease compaction time by 3x) by @codetheweb in #2736
• [ENH]: purge block cache after compaction by @codetheweb in #2741
• [PERF] Move write to blockfiles off main runtime by @sanketkedia in #2740
• [ENH] Synchronize post s3.get() operations in providers by @sanketkedia in #2712
• [ENH]: make test_sanity recall based by @codetheweb in #2742
• [PERF] Add include metadata to MetadataReader to avoid unncessary payload size by @HammadB in #2750
• [PERF]: speed up get_target_block_id() by @codetheweb in #2743
• [ENH]: skip querying MetadataSegmentReader for empty where clause by @codetheweb in #2711
• [ENH] reset back to previous JS api, keeping other improvements by @jeffchuber in #2714
• [BLD]: add debug info to Rust release binaries by @codetheweb in #2773
• [ENH] Weighted LRU for HNSW provider cache by @Ishiihara in #2697
• [ENH]: use jemalloc for compactor service by @codetheweb in #2777
• [BUG] Smaller batch size by @HammadB in #2781
• [BLD]: bump version of upload-artifact actions by @codetheweb in #2782
• [DOC] fix meta tags by @jeffchuber in #2786
• [BLD]: use cache mount in Rust Dockerfile by @codetheweb in #2779
• [ENH]: GET /databases should return 404 instead of 500 for not found by @codetheweb in #2788
• [BLD]: run Rust integration tests in CI by @codetheweb in #2787
• [CLN] Propagate errors for block get/fork by @HammadB in #2778
• [DOC] fix docs build by @jeffchuber in #2794
• [DOC] Add description for query_uris parameter in collection query method by @ragul-kachiappan-dev in #2656
• [DOC] add levands clojure client to docs by @jeffchuber in #2802
• [DOC] Update JS client documentation by @david1542 in #2233
• [DOC]: use pnpm add <pkg> instead of pnpm install <pkg> by @jferrettiboke in #2629
• [DOC] Update the add function in the guide, which is currently using the old version with text instead of documents. by @sonnhfit in #2550
• [CHORE] DeprecationWarning in httpx_content.py:202: by @LukasWestholt in #2509
• [ENH] add docs for vacuum CLI command by @codetheweb in #2595
• [DOC]: remove docs for reverted JS refactor by @codetheweb in #2805
• [DOC]: Update Langchain integration example and fix broken link by @ishaksebsib in #2771
• [DOC]: Yarn package install in getting-started.md by @DuYongLee in #2566
• [BUG] - Fix tailwind themes in docs by @itaismith in #2808
• [CLN] Support numpy >=2.0 by @HammadB in #2811
• [BUG] Test that query result shapes are correct in invariants by @drewkim in #2807
• [CLN] Add batch size errors by @HammadB in #2812
• [BUG] Fix CVP tests by forcing numpy reimport by @HammadB in #2814
• [BUG] Fix I32Array BW compat by @HammadB in #2815
• [ENH] Update client telemetry for developer cloud by @itaismith in #2785
• [RELEASE] 0.5.6 by @itaismith in #2817

New Contributors

• @ragul-kachiappan-dev made their first contribution in #2656
• @david1542 made their first contribution in #2233
• @jferrettiboke made their first contribution in #2629
• @sonnhfit made their first contribution in #2550
• @LukasWestholt made their first contribution in #2509
• @ishaksebsib made their first contribution in #2771
• @DuYongLee made their first contribution in #2566
• @itaismith made their first contribution in #2808
• @drewkim made their first contribution in #2807

Full Changelog: 0.5.5...0.5.6