v0.7.0

What's Changed

• Add min/max implementations for L0 path by @kurapov-peter in #487
• 324 replace the llvm legacy pass manager with the llvm new pass manager clean by @lmontigny in #417
• Add gpu hang bucket by @kurapov-peter in #492
• Bump to v0.7.0dev by @alexbaden in #497
• Update googlebenchmark to v1.8.0 by @alexbaden in #500
• Missing atomic header to compile without conda env by @lmontigny in #493
• Add access to log options in PyHDK. by @ienkovich in #494
• Avoid overflow in get_matching_group_value. by @ienkovich in #495
• Fixed Calcite processing of queries with APPROX_QUANTILE aggregation. by @AndreyPavlenko in #506
• Enable multifrag result set tables by default. by @ienkovich in #502
• Use multifrag kernels on CPU for groupby with a big result. by @ienkovich in #505
• [Python build] Reduce warnings by @Devjiu in #503
• Added sanity tests run on windows by @gshimansky in #486
• Add window functions support in QueryBuilder. by @ienkovich in #508
• Add mandatory getFieldName for all nodes. by @ienkovich in #510
• Add missing corr bindings. by @ienkovich in #509
• Avoid null comparison for non-nullable args. by @ienkovich in #514
• Relaxed string dictionary fallback constraint for L0 by @akroviakov in #519
• Fix conda-forge build by @leshikus in #511
• Add docker clang build by @leshikus in #512
• [Time Conversion] Timestamp to time by @Devjiu in #482
• Fixed window functions json deserialization in Calcite by @AndreyPavlenko in #526
• Fixed timestamp literals serialization. by @AndreyPavlenko in #457
• Use the correct Dockerfile in L0 test builds by @alexbaden in #532
• Avoid uint32_t for column offset. by @ienkovich in #530
• Fix cmake whitespace warning by @alexbaden in #531
• Cast slot index to match slot offset for columnar output by @alexbaden in #533
• Update guava and add log ignore by @alexbaden in #536
• Require TBB for CPU multithreading by @alexbaden in #539
• Move window function pre-project creation to DAG canonicalization. by @ienkovich in #545
• Support arrow conversion for multi-fragment result set tables [1/N] by @ienkovich in #546
• Add lazy dict materialization to ConfigBuilder by @alexbaden in #550
• Allow custom Calcite (java) / HDK log paths by @alexbaden in #524
• Enable shared memory by default by @kurapov-peter in #534
• Add execution mode option to PyHDK python tests by @kurapov-peter in #518
• Move TableFragmentsInfo and FragmentInfo code to proper lib. by @ienkovich in #554
• Support arrow conversion for multi-fragment result set tables [2/N] by @ienkovich in #548
• Fix cleanup by @leshikus in #551
• Add cpu-only testing for L0-enabled build by @kurapov-peter in #515
• Add non-lazy data import. by @ienkovich in #553
• Support partitioned aggregation. by @ienkovich in #538
• Add missing destructor by @kurapov-peter in #556
• Support arrow conversion for multi-fragment result set tables [3/3] by @ienkovich in #555

New Contributors

• @AndreyPavlenko made their first contribution in #506

Full Changelog: v0.6.0...v0.7.0