Latest

Version: 1.0.16.dev60
Git ref: refs/heads/main
Build Date: 2025-07-18T15:56
PIP Package: chroma-1.0.16.dev60.tar.gz
Github Container Registry Image: :1.0.16.dev60
DockerHub Image: :1.0.16.dev60

cli-1.1.4

CLI release.

1.0.15

Version: 1.0.15
Git ref: refs/tags/1.0.15
Build Date: 2025-07-02T17:07
PIP Package: chroma-1.0.15.tar.gz
Github Container Registry Image: :1.0.15
DockerHub Image: :1.0.15

What's Changed

• [BUG] Track the threshold of garbage collected fragments by @rescrv in #4874
• [ENH] Make the Failed to Fetch log error verbose by @rescrv in #4886
• [ENH] Put a limiter on the purge_dirty_for_collection call. by @rescrv in #4887
• [BUG] correctly remove embeddings, embeddings metadata, and segment metadata on delete_collection by @jairad26 in #4884
• [BUG] Contention in S3 gets assumed to be retryable, leading the manifest to fail. by @rescrv in #4888
• [RELEASE] JS 3.0.4 by @itaismith in #4890
• [ENH] A scrubbing tool that supports limits. by @rescrv in #4891
• [BUG]: fix duplicate DeleteUnusedFiles task in GC for soft-deleted collections by @codetheweb in #4892
• [ENH] Add a path to bail from snapshot, manifest, and gc installs. by @rescrv in #4893
• [ENH]: make S3 tracing spans less verbose by default by @codetheweb in #4861
• [CLN] Remove a spammy log line about Scouted Logs from the Go log by @rescrv in #4895
• [BUG] Panic on sysdb when calling CheckCollection by @rescrv in #4899
• [BUG] An transient error in scrubbing leads to scrub error. by @rescrv in #4898
• [ENH]: Return database id in get collections call from sysdb by @sanketkedia in #4686
• [ENH][wal3] Make GC be delete-free for log/gc/GARBAGE by @rescrv in #4904
• [BUG] Manifest-initial-offset was not set under gc conditions. by @rescrv in #4903
• [ENH] CLI - Set Chroma env variables by @itaismith in #4711
• [RELEASE] CLI 1.1.3, JS 3.0.5, Python 1.0.14 by @itaismith in #4913
• [ENH]: Plumb prefix path all the way to the bf writer by @sanketkedia in #4743
• [CHORE] Add prettier, fix build & tests for new-js client by @jairad26 in #4914
• [BUG] The condition for setting the manifest initial_seq_no was broken. by @rescrv in #4920
• [ENH]: Support writing data to separate prefixes in s3 by @sanketkedia in #4753
• [BUG] Fix a flakey prop test, and commit the regression. by @rescrv in #4922
• [BLD] Update deploy automation to deploy control and data planes separately by @eculver in #4917
• [DOC] Regex docs by @itaismith in #4923
• [ENH]: enforce maximum get_collections limit as 100 by @c-gamble in #4929
• [ENH] Purge dirty log in background at the end of scheduled compaction by @Sicheng-Pan in #4915
• [ENH] Move Log GC to operator by @Sicheng-Pan in #4919
• [BUG] Do not leak tokio tasks in the log service. by @rescrv in #4936
• [BUG] Log GC offset should be one above minimum compaction offset by @Sicheng-Pan in #4938
• [ENH] Make roll dirty log always converge to coalesce everything. by @rescrv in #4927
• [ENH] Read args from env variables for Python CloudClient by @itaismith in #4939
• [TST] Add tests for CheckCollections, make test collection setup more flexible by @eculver in #4912
• [DOC] Add cloud docs, architecture docs, data model doc by @HammadB in #4859
• [BUG] Coalesce when multiple collections return the same info to compact by @rescrv in #4946
• [BUG] Enrich from the manifest if a cursor doesn't exist. by @rescrv in #4947
• [CHORE]: bump default regex quota to 256 by @c-gamble in #4949
• [ENH] Allow slicing of the log when pulling to narrow down problems. by @rescrv in #4925
• [ENH]: More concurrent blockfilewriter by @sanketkedia in #4889
• [ENH] If the dirty log fails with LogContentionDurable, do not fail the operation. by @rescrv in #4953
• [ENH] Warn, not error, if dirty log has no cursor. by @rescrv in #4952
• [ENH]: use TracedJson for /upsert & /update by @codetheweb in #4956
• [ENH]: granular lock for posting list by @sanketkedia in #4937
• [BUG] Read from legacy metadata config when no collection config set by @jairad26 in #4945
• [ENH][wal3] Add tracing to know how many iters do_work does. by @rescrv in #4958
• [PERF] increase delta conversion speed by batching by @HammadB in #4551
• [ENH] Cancellation safety for append_batch. by @rescrv in #4959
• [BUG] Fix forking for js client by @itaismith in #4962
• [BUG] Be more resilient to failure. by @rescrv in #4964
• [CHORE]: allow authz resources and actions to be serializable by @c-gamble in #4960
• [CHORE] pin posthog dependency <6.0.0 to account for breaking change by @jairad26 in #4970
• [BUG][wal3] GC gets wedged by @rescrv in #4972
• [ENH]: improve ListCollectionsToGc with filter for min alive versions by @codetheweb in #4971
• [ENH] Add robots.txt to allow all web crawlers on docs.trychroma.com by @philipithomas in #4940
• [ENH] Skip log GC in dry run mode by @Sicheng-Pan in #4977
• [ENH]: Add ability to set different block sizes for different blockfiles by @sanketkedia in #4948
• [PERF] Make IO accesses parallel by @sanketkedia in #4988
• [PERF]: Skip creating readers if filter is null by @sanketkedia in #4989
• [TST] Test for #4972 by @rescrv in #4983
• [CLN] Fix dedup in get_collections_with_new_data. by @rescrv in #4974
• [ENH] Implement three-phase garbage collection for WAL3 by @rescrv in #4984
• [CLN] Remove err(Display) from wal3. by @rescrv in #4992
• [BLD] Optimize which tests we run on Windows by @eculver in #4955
• [CHORE]: use a match for defining default quotas by @c-gamble in #4941
• [BUG] Batch inserts on push_logs in sqlite by @jairad26 in #4982
• [ENH] Add config to disable log GC by @Sicheng-Pan in #5002
• [ENH] Wire up garbage collector to do 3-phase GC. by @rescrv in #4987
• [CHORE] Fix pr title workflow to allow edits by @jairad26 in #5001
• [ENH] Do not materialize all fragments to delete. by @rescrv in #5004
• [ENH] Better JS client error messaging by @kylediaz in #4944
• [ENH] Pipeline compactions for different collections by @tanujnay112 in #4963
• [DOC] IA refresh by @itaismith in #4928
• [BUG]: Spann indexing bug by @sanketkedia in #5011
• [ENH]: migrate current metering functionality to new metering lib by @c-gamble in #4868
• [CLN] Add max_batch_size to sqlite_log, use in push_logs by @jairad26 in #5010
• [ENH]: Upgrade foyer to 0.17.3 by @sanketkedia in #5013
• [ENH]: add request timing to metering by @c-gamble in #4877
• [RELEASE] CLI 1.1.4, JS 3.0.6, Python 1.0.15 by @itaismith in #5015

Full Changelog: 1.0.13...1.0.15

cli-1.1.3

CLI release.

1.0.13

Version: 1.0.13
Git ref: refs/tags/1.0.13
Build Date: 2025-06-18T22:16
PIP Package: chroma-1.0.13.tar.gz
Github Container Registry Image: :1.0.13
DockerHub Image: :1.0.13

What's Changed

• [PERF]: skip document mutation in full-text index writer if old and new documents are same by @codetheweb in #4678
• [BUG] Rust log service fork offsets off by one by @Sicheng-Pan in #4689
• [ENH] Use tracing util to set root span on rust log service. by @rescrv in #4691
• [ENH]New JS Client by @itaismith in #4632
• [CHORE] Fix Docker publish action by @itaismith in #4697
• [CHORE] Instrumentation of dirty log rollup by @rescrv in #4698
• [ENH] Refactor dirty log rollup for async implementation by @rescrv in #4700
• [BUG]: correctly update collections.num_versions upon compaction flush by @codetheweb in #4690
• [CHORE] unpin fastapi, move to dev dependency by @jairad26 in #4706
• [CHORE] Add pathlib support to path for persistence by @jairad26 in #4709
• [TST] More tests of the log service code. by @rescrv in #4707
• [BUG] Fix where validation bug by @itaismith in #4715
• [CHORE] Log frag count during scout by @Sicheng-Pan in #4714
• [ENH]: add 2 params to GetCollections: include soft deleted & collection IDs by @codetheweb in #4717
• [ENH] Cache manifests that get used for pull logs. by @rescrv in #4726
• [ENH] Avoid LogContention on freshly sealed logs. by @rescrv in #4724
• [TST] Unit test log forking under migration by @Sicheng-Pan in #4729
• [ENH] Add nac delay histogram metric by @HammadB in #4731
• [ENH] Mark a fork as dirty if it has records on the log. by @rescrv in #4732
• [DOC] - Fixes broken URLs by @tjkrusinskichroma in #4740
• [DOC] - Fixes various css isssues by @tjkrusinskichroma in #4738
• [DOC] Docs for new JS client by @itaismith in #4741
• [BUG] Create a bunch of collections and delete them and the dirty log will not roll up. by @rescrv in #4727
• [ENH] Avoid loading the manifest and rely exclusively on dirty log for enrich logs by @rescrv in #4725
• [ENH] Add getCollectionById to JS by @itaismith in #4744
• [DOC] Update EF docs for new JS by @itaismith in #4749
• [ENH] Simplify forking logic in go log by @rescrv in #4734
• [BUG] Fix a bug in the wal3 records-based pruning. by @rescrv in #4750
• Revert "[ENH] Simplify forking logic in go log" by @rescrv in #4754
• [BUG] removed the masked err from ForkLogs by @rescrv in #4739
• [BLD]: fix Tilt setup flakes by @codetheweb in #4720
• [BLD]: build arm64 image on arm runner instead of in QEMU by @codetheweb in #4759
• [CHORE] Add tracing for continuous verification by @jasonvigil in #4756
• [BUG] Make sure the cached manifest can satisfy a pull-logs request. by @rescrv in #4758
• [BLD]: split Tilt setup & cluster bring up into separate actions by @codetheweb in #4762
• [BLD]: specify working directory in reusable Tilt setup action by @codetheweb in #4763
• [ENH] Enable wal3 for default tenant. by @rescrv in #4761
• [BUG] Seal collections on fork in the go log. by @rescrv in #4767
• [BUG] Keep TempDir obj for test by @Sicheng-Pan in #4755
• [TST] Implement proptest for rust log service by @Sicheng-Pan in #4764
• [BUG]: properly escape _ and escape % rather than stripping for SQLite $contains filter by @codetheweb in #4402
• [ENH]: GCv2: add grace period for transitioning soft deleted collections -> hard deleted by @codetheweb in #4719
• [BLD] [REVERT]: build arm64 image on arm runner instead of in QEMU by @codetheweb in #4773
• [TST]: fix concurrency bug in test_cross_version_persist by @codetheweb in #4770
• [DOC] - Support mobile on docs.trychroma.com by @tjkrusinskichroma in #4766
• [BUG] Fix an obo in the wal3 reader. by @rescrv in #4772
• [CHORE] Build fixes for new js client by @itaismith in #4752
• [TST]: fix test filtering flake caused by wildcard handling change by @codetheweb in #4775
• [HOTFIX][BUG] Fix an obo in the wal3 reader. (#4772) by @rescrv in #4777
• [BLD]: fix Go test flakes by @codetheweb in #4760
• [BLD]: speed up Rust tests in CI by @codetheweb in #4780
• [ENH]: speed up filter benchmark by @codetheweb in #4782
• [TST] More proptest for rust log service by @Sicheng-Pan in #4771
• [ENH] Log client healthcheck by @HammadB in #4783
• [BLD]: shard Rust integration tests for speedup by @codetheweb in #4676
• [ENH] Set num_records_before_backpressure and test it. by @rescrv in #4769
• [BUG] Populate client args from deprecated args by @itaismith in #4786
• [TST]: fix flakes in Rust tests caused by port conflicts by @codetheweb in #4788
• [BLD]: speed up test_add.py against cluster by @codetheweb in #4781
• [BLD]: remove usage of forked Blacksmith actions by @codetheweb in #4789
• [TST] Move windows-based tests out of PR Checks by @eculver in #4748
• [ENH] Tool to purge a collection from the dirty log. by @rescrv in #4787
• [ENH] Remove CoreML as a provider for default ef by @jairad26 in #4779
• [BLD]: move GC proptest into separate job by @codetheweb in #4794
• [CLN] Unify operator types by @Sicheng-Pan in #4656
• [BUG] Fix test parallelization by @eculver in #4797
• [DOC] Add missing quotes by @BrodaNoel in #4795
• [CHORE] address logger deprecation warnings by @emmanuel-ferdman in #4668
• [ENH] Tool to inspect the contents of the log. by @rescrv in #4757
• [BUG] Clean up tracing by @drewkim in #4793
• [ENH] Add resource_name to SysDB tenants table by @drewkim in #4730
• [TST]: fix local SQLite fixtures: every test should have a unique persistent directory by @codetheweb in #4800
• [ENH] Add Copy API to chroma storage. by @rescrv in #4798
• [ENH] wal3::copy implemented using scan/AWS copy by @rescrv in #4803
• [TST]: speed up test_embeddings.py against cluster by @codetheweb in #4796
• [BLD]: use sccache during PyPi build by @codetheweb in #4806
• [TST]: fix test_rebuild flake by @codetheweb in #4808
• [DOC] Update single node performance benchmarks by @jairad26 in #4695
• [ENH] Instrument s3 copy and delete by @Sicheng-Pan in #4811
• [BLD]: pin Python version for wheel build workflow by @codetheweb in #4815
• [BLD]: remove architecture specification from actions/setup-python by @codetheweb in #4816
• [TST] Impl proptest for log seal and migrate by @Sicheng-Pan in #4810
• [DOC] Fix link in embedding docs by @kylediaz in #4805
• [BLD]: disable sccache during maturin build by @codetheweb in #4819
• [ENH]: bump AWS Go S3 SDK to v2 by @codetheweb in #4820
• [ENH]: improve garbage collector tracing by @code...

1.0.12

Version: 1.0.12
Git ref: refs/tags/1.0.12
Build Date: 2025-05-31T02:41
PIP Package: chroma-1.0.12.tar.gz
Github Container Registry Image: :1.0.12
DockerHub Image: :1.0.12

What's Changed

• [ENH] Add rust log service memberlist by @HammadB in #4415
• [BUG] Wire up the rust-log-service member ID to the dirty log. by @rescrv in #4645
• [CLN] Failed precondition -> Internal by @HammadB in #4646
• [BUG] Bump testing back to one rust log service replica. by @rescrv in #4648
• [ENH][mdac] Demonstrate and test how to give per-tenant exclusions. by @rescrv in #4639
• [BUG]: don't create duplicate segments for collection under concurrent get_or_create requests by @codetheweb in #4647
• [BUG] Use the right path for my_member_id in rust-log-service. by @rescrv in #4652
• [DOC] fix docs for list_collections by @jairad26 in #4649
• [ENH] Cleanup the single most spammy log line in rls. by @rescrv in #4654
• [ENH]: sysdb changes to support moving collection hard deletes to garbage collector by @codetheweb in #4607
• [ENH]: serde util for optional u128 by @c-gamble in #4659
• [BUG]: root collection ID should be null if empty string by @codetheweb in #4663
• [ENH] Clean up client manager into manager/assigner - make log client use it by @HammadB in #4640
• [ENH] Make explicit seal/migrate calls for the log service. by @rescrv in #4669
• [ENH]: add readiness probe for garbage collector by @codetheweb in #4662
• [ENH]: perform collection hard deletes from garbage collector by @codetheweb in #4605
• [TST] Regex testing by @Sicheng-Pan in #4570
• [ENH]: soft delete databases, add FinishDatabaseDeletion gRPC method to hard delete databases by @codetheweb in #4627
• [BLD]: fix typo so that test durations are reported properly by @codetheweb in #4674
• [ENH]: wire GC v2 to new cleanup modes & call FinishDatabaseDeletion from garbage collector by @codetheweb in #4671
• [ENH] Robustly handle errors on the compaction path of the log. by @rescrv in #4677
• [BUG] Raise Error when can't deserialize configuration json from server, lazily load ef on CollectionModel, warn on api_key by @jairad26 in #4471
• [ENH] Add logging statements to figure out why forking is failing in staging. by @rescrv in #4679
• [ENH] Prefetch block by prefixes by @Sicheng-Pan in #4623
• [ENH] Add validation when multiple embedding functions set on client by @jairad26 in #4507
• [BLD]: parallelize cross version test by @codetheweb in #4675
• [ENH]: don't skip GC for collection in fork tree if using GC v2 by @codetheweb in #4681
• [ENH] add Mistral embedding function, fix some naming mismatches across clients by @jairad26 in #4653
• [ENH][chroma-load] Allow reference data sets to reference minilm6v2. by @rescrv in #4687
• [RELEASE] 1.0.12 by @jairad26 in #4685

New Contributors

• @c-gamble made their first contribution in #4659

Full Changelog: 1.0.11...1.0.12

1.0.10

Version: 1.0.10
Git ref: refs/tags/1.0.10
Build Date: 2025-05-22T03:00
PIP Package: chroma-1.0.10.tar.gz
Github Container Registry Image: :1.0.10
DockerHub Image: :1.0.10

What's Changed

• [CHORE] Use bigger machine for build and publish container image by @HammadB in #4533
• [ENH] Expose may_contain for disk cache, use in prefetch by @HammadB in #4530
• [BUG]: use byte offset in full-text reader rather than token position by @codetheweb in #4531
• [ENH]: Add spann metrics by @sanketkedia in #4492
• [ENH]: Add name to various foyer caches by @sanketkedia in #4532
• [ENH]: ListCollectionsToGc returns lineage file path, group by fork tree, accept optional tenant for filtering by @codetheweb in #4529
• [BUG]: increase max payload size of log service (Go) by @codetheweb in #4534
• [ENH] Fix 422 Unprocessable Entity error in Chroma Load by @rescrv in #4537
• [CHORE] Reduce default NUM_FORKS quota rule value by @drewkim in #4525
• Revert "[BUG]: increase max payload size of log service (Go) (#4534)" by @eculver in #4540
• [CHORE] Fix test-embeddings.py and test_sanity by @rescrv in #4526
• [BUG] Default embedding function in JS did not create the correct config dict by @jairad26 in #4539
• [ENH] Emit log_uncompacted_record_count from rust log service. by @rescrv in #4541
• [DBG] Add temporary trace for filter argument by @Sicheng-Pan in #4548
• [ENH] Support custom data sets for chroma load. by @rescrv in #4547
• [BUG] Restore increase max payload size of log service (Go) by @eculver in #4546
• [BUG]: v0 in version file for forked collection missing segment file paths from source collection by @codetheweb in #4550
• [ENH] Add log sealing to the go service. by @rescrv in #4554
• [BUG]: Regex search returning documents that are deleted by @sanketkedia in #4562
• [BUG] Add hosted-frontend to OTel stdout layer by @drewkim in #4564
• [BUG]: Handle version file creation with empty file paths by @sanketkedia in #4552
• [BLD] Do not trigger deploy on merge to release branches by @eculver in #4561
• [ENH] For local: use subquery for FTS, unions for int & float metadata expr, is true -> is not null by @jairad26 in #4556
• [ENH] Add a safety cutoff to the rust log service. by @rescrv in #4535
• [ENH] Add handling for ChromaQuotaExceededError by @gshahbazian in #4569
• [ENH] bump JS client to v2.4.5 by @philipithomas in #4574
• [CHORE] Remove validation on ef config update for model path by @jairad26 in #4571
• [ENH] Add a tool to reason through the state space of bootstrap. by @rescrv in #4558
• [RELEASE] CLI 1.1.2 by @itaismith in #4572
• [ENH] Bootstrap a wal3 log from existing content. by @rescrv in #4560
• [BUG] Fix race in get_or_create by @HammadB in #4568
• [BLD]: use blacksmith pull-through container image cache for Tilt cluster by @codetheweb in #4542
• [ENH]: SysDb should return lineage, version file paths and root collection ID on collections by @codetheweb in #4557
• [BLD]: use uv in CI by @codetheweb in #4581
• [BLD] retry kind setup by @codetheweb in #4584
• [RELEASE] 1.0.10 by @itaismith in #4587

Full Changelog: 1.0.9...1.0.10

cli-1.1.2

CLI release.

1.0.9

Version: 1.0.9
Git ref: refs/tags/1.0.9
Build Date: 2025-05-13T03:39
PIP Package: chroma-1.0.9.tar.gz
Github Container Registry Image: :1.0.9
DockerHub Image: :1.0.9

What's Changed

• [CHORE] Fix JS release to support platform specific bindings by @itaismith in #4442
• [RELEASE] Release JS client 2.4.1 by @itaismith in #4443
• [RELEASE] JS Client 2.4.2 by @itaismith in #4444
• [CLN] Warn on log pull failure instead of error by @HammadB in #4435
• [RELEASE] JS Client 2.4.3 by @itaismith in #4448
• [TST] Debug cli tests by @itaismith in #4453
• [ENH] replace group_by and having with distinct and where for local chroma by @jairad26 in #4428
• [CHORE] Update JS bindings CI by @itaismith in #4445
• [ENH] add example of forking using github repos by @jairad26 in #4413
• [DOC] Add docs for CLI-1.1.0 by @itaismith in #4440
• [DOC] Add JS install for the CLI by @itaismith in #4459
• [ENH]: Add retry for add, update and upsert by @sanketkedia in #4457
• [ENH] Add load service to local dev tiltfile by @jasonvigil in #4397
• [CLN] Pass Chroma API key via env var for chroma-load-start by @jasonvigil in #4469
• [ENH] A route and tool to inspect the dirty log. by @rescrv in #4461
• [ENH] Change rust-log-service to a stateful set. by @rescrv in #4477
• [ENH]: Introduce NUM_REGEX_PREDICATES by @sanketkedia in #4429
• [CLN] Cleanup a warning in js_bindings/Cargo.toml. by @rescrv in #4481
• [BUG] Remove misleading log message by @jasonvigil in #4480
• [ENH] Add caching to rust log service. by @rescrv in #4465
• [BUG] Fix CLI update command by @itaismith in #4482
• [ENH]: When SPANN is enabled, route HNSW configuration to SPANN, remove enable_set_index_params by @sanketkedia in #4470
• [ENH] Add RPC on SysDB to get fork count for a collection by @drewkim in #4484
• [BUG] Retry chroma-load upserts when rate limited by @jasonvigil in #4485
• [CLN] CLI - Refactor webbrowser commands by @itaismith in #4455
• [BUG] Restore commented-out test. by @rescrv in #4478
• [ENH][chroma-load] If no api key is present in env, do not auth. by @rescrv in #4486
• [ENH] Configure the cache to have a hostPath and mountPath for log service. by @rescrv in #4483
• [ENH]: Unique constraint violation when two writes contend for collection table on log by @sanketkedia in #4494
• [ENH] write embedding function to config if provided by @jairad26 in #4496
• [ENH] Disallow empty string ids during add by @jairad26 in #4488
• [BUG] Adjust the ping-pong test for wal3 to timeout longer and run less. by @rescrv in #4495
• [ENH] Allow collections to shunt to the alt log by tenant. by @rescrv in #4498
• [ENH][rust-log-service] Use the persistent cache. by @rescrv in #4499
• [CHORE] Support count_forks/num_forks in TestSysDB/Quotas by @drewkim in #4497
• [ENH] Display errors for all log calls. by @rescrv in #4504
• [ENH]: Add quota on regex pattern length by @sanketkedia in #4503
• [BUG] Bump buggy deps to satisfy dependabot by @rescrv in #4505
• [ENH] Make metadata optional to allow deleting metadata in python client by @jairad26 in #4490
• [BLD]: unify Rust Dockerfiles for faster builds by @codetheweb in #4479
• [ENH] QuotaExceededError can include an optional message by @philipithomas in #4489
• [RELEASE] CLI 1.1.1 by @itaismith in #4509
• [CLN] fork_collection quota cleanup by @drewkim in #4508
• [BLD]: fix secret usage for Go test reusable workflow by @codetheweb in #4510
• [ENH] Implement literal expression for regex by @Sicheng-Pan in #4421
• [CHORE] Reimplement fork backstop by @drewkim in #4511
• [ENH] Implement literal provider by @Sicheng-Pan in #4430
• [ENH] Wireup regex in filter operator by @Sicheng-Pan in #4452
• [BUG] Fix blockfile range scan by @Sicheng-Pan in #4467
• [CHORE] Add a log line to go server to see if its scout logs is called by @rescrv in #4514
• [CLN] Rename $matches to $regex by @Sicheng-Pan in #4506
• [ENH] Reorganize load-service traces by @jasonvigil in #4518
• [RELEASE] 1.0.9 by @HammadB in #4523

Full Changelog: 1.0.8...1.0.9

1.0.8

Version: 1.0.8
Git ref: refs/tags/1.0.8
Build Date: 2025-05-05T09:26
PIP Package: chroma-1.0.8.tar.gz
Github Container Registry Image: :1.0.8
DockerHub Image: :1.0.8

What's Changed

• [ENH] add collection forking to js client by @philipithomas in #4372
• [BUG] Fix race between hnsw load and hnsw purge by @sanketkedia in #4345
• [RELEASE] js client v2.3.0 with collection config and forking by @philipithomas in #4373
• [ENH] Make connection configurable per running load-service workload by @jasonvigil in #4374
• [BUG] Handle uninitialized spann segment reader by @sanketkedia in #4348
• [TST] Add property test for collection forking by @Sicheng-Pan in #4375
• [ENH] Fixed Dockerfile for the current genertion of chroma project by @j0hnthebuilder in #4270
• [BUG] Ann calculation in tests after collection config + handle none in collection config by @sanketkedia in #4350
• [ENH] Turn on spann by default by @sanketkedia in #4351
• [CHORE] Fix flaky default_ef test by @jairad26 in #4376
• [CLN]: Move hnsw_provider.open() lock inside the call itself by @sanketkedia in #4383
• [CHORE] Move distributed to default collection configuration by @jairad26 in #4377
• [CHORE] Fix type safety by @jairad26 in #4339
• [ENH] Add Cloudflare Worker AI Embedding Function by @jairad26 in #4389
• [BUG]: Fix Voyage AI EF to match new API by @tazarov in #4392
• [DOC]: Fix wrong cohere import in collection config docs by @tazarov in #4393
• [ENH] Track collection forking events by @Sicheng-Pan in #4395
• [CHORE] remove hnsw-lib from core dependencies by @jairad26 in #4385
• [ENH] Add python & js client support to query on subset of IDs by @jairad26 in #4250
• [ENH]: Huggingface Embedding Server authorization by @MrExplode in #4227
• [ENH]: better errors in ComputeUnusedFilesOperator by @codetheweb in #4400
• [ENH] Add Together AI Embedding function in Python and JS by @jairad26 in #4394
• [BUG]: don't try reading HNSW files as sparse indices when performing garbage collection on SPANN collections by @codetheweb in #4401
• [BUG] Prevent unconditional unwrap of collection config by @jasonvigil in #4404
• [BUG]: fix missing tenant ID in version file by @codetheweb in #4408
• [BUG] Make a strongly-consistent get operation. by @rescrv in #4407
• [BUG] Tag prefetch record as IO, fix spinloop in admissionController by @HammadB in #4405
• [TST] Re-enable parallel spann tests with bigger stack size by @sanketkedia in #4409
• [ENH] Wire up regex filter from client to query node by @Sicheng-Pan in #4410
• [CLN] Fix Port Parameter Inconsistency in HttpClient (String to Integer) by @iw4p in #3662
• [BUG] Fix the boundaries for snapshots. by @rescrv in #4411
• [CLN] Remove legacy python FE from tilt by @HammadB in #4414
• [ENH] Fork the wal3 log. by @rescrv in #4416
• [ENH] Enable rust log service in CI and add some tracing. by @rescrv in #4418
• [ENH]: Enable auth for forking by @sanketkedia in #4419
• [BUG] You must ask for strong consistency from S3 GET. by @rescrv in #4420
• [ENH] CLI client refactor by @itaismith in #4417
• [BUG] Handle duplicates in chroma-load. by @rescrv in #4423
• [BUG] ignore trigrams with null terminator byte when constructing full text index by @codetheweb in #4422
• [ENH] Restore chroma-load's random-upsert ability for Reference datasets. by @rescrv in #4425
• [CLN] Use GRPC for chroma-load OTEL by @jasonvigil in #4427
• [ENH] Add support to convert Get/QueryResult to pandas dataframe by @jairad26 in #4304
• [ENH] CLI Browse command by @itaismith in #4431
• [RELEASE] Release CLI 1.1.0 by @itaismith in #4438
• [RELEASE] JS release 2.4.0 by @itaismith in #4439
• [RELEASE] 1.0.8 by @itaismith in #4441

New Contributors

• @j0hnthebuilder made their first contribution in #4270
• @MrExplode made their first contribution in #4227
• @iw4p made their first contribution in #3662

Full Changelog: 1.0.7...1.0.8