Release v0.30.0 · lancedb/lance

What's Changed

feat(java): support checkout version by @yanghua in #3907
feat: wrap FilterExec in LanceFilterExec by @wkalt in #3939
feat: add storage options for LanceFileReader by @leaves12138 in #3941
feat: prefix matching for FTS by @BubbleCal in #3935
feat(java): support list versions and checkout version in Dataset by @majin1102 in #3945
feat: add a few lance examples in Rust by @HaochengLIU in #3817
feat: rebase conflict between rewrite and frag reuse index cleanup by @jackye1995 in #3952
feat(java): support tag operations on Java dataset by @majin1102 in #3958
feat(java): support restoring to the checked out version by @majin1102 in #3969
feat: expose update_config api for python by @yanghua in #3963
feat: support returning tags in order of version with descending order as default by @Jay-ju in #4001
feat: add support for must_not clause in Boolean queries by @BubbleCal in #4012
feat: support unenforced primary key concept in schema by @jackye1995 in #4002
feat: redirect Spark documentation to new lance-spark website by @jackye1995 in #4032
feat: support enable/disable auto-cleanup by @yanghua in #4037
feat: add cache stats for new cache by @wjones127 in #4024

fix: correctly specify schema in encoded substrait for expr by @westonpace in #3937
fix: newer indices may be read from cache by @BubbleCal in #3936
fix: null value rows missing after bitmap remap by @jackye1995 in #3956
fix: rebase transactions from oldest to latest by @jackye1995 in #3955
fix: typos in documentation comments by @leopardracer in #3970
fix: call remove_outer_validity() before dictionary_encode by @luohao in #3974
fix: don't crash when reading index trained on old version by @westonpace in #3978
fix: handle case in 2.1 writer where lists are masked by null struct by @westonpace in #3946
fix: don't panic on DF table provider if test has filter on indexed column by @westonpace in #3954
fix: respect precomputed partitions and shuffle buffers for v3 index by @BubbleCal in #3910
fix: panic caused by reading empty posting list by @BubbleCal in #4019
fix: optimize index with no new data by @wjones127 in #4017
fix: filter out null row ids by @BubbleCal in #4020
fix: separate loading configuration handling for jieba and lindera by @KazuhitoT in #3932
fix: reading large fragment reuse index details fails with file not found by @jackye1995 in #4035
fix: use i32 for dictionary encoding indices by @luohao in #3898
fix: optimizing FTS causes error if it's without positions by @BubbleCal in #4044

refactor: change the default process-wide IOPS limit to be 128 by @westonpace in #3972

Full Changelog: v0.29.0...v0.30.0