v0.6.0

What's Changed

• Fix python version to 3.9 by @leshikus in #328
• L0 + CUDA build by @kurapov-peter in #329
• [PyHDK] Forwarding import of parquet files from cpp by @Devjiu in #292
• Use __atomic_compare_exchange_n for CPU CAS atomics where intrinsics … by @alexbaden in #338
• Add logging option to hdk init by @kurapov-peter in #339
• Bump cmake version to 0.5.0 by @kurapov-peter in #344
• Add ResultSetRegistry storage [1/N] by @ienkovich in #331
• Revert pr266 by @leshikus in #341
• Remove ExecutionResult::getDataPtr. by @ienkovich in #350
• Update RELEASE.md by @alexbaden in #352
• Remove glog dependency by @alexbaden in #349
• Bump to v0.6.0dev by @alexbaden in #357
• Add nightly conda-forge build by @leshikus in #356
• Add ResultSetRegistry storage [2/N] by @ienkovich in #348
• Fix table reordering for self-join case. by @ienkovich in #364
• Move RowSetMemoryOwner code from QueryEngine. by @ienkovich in #353
• Support non-encoded strings in Arrow converter. by @ienkovich in #359
• Add an issue bucketing script for query execution tests by @kurapov-peter in #368
• Fix cuda by @leshikus in #371
• Fix docker cache by @leshikus in #369
• Fallback to CPU in case code generation fails. by @kurapov-peter in #373
• hdk_cas fixes for CUDA and Windows by @ienkovich in #375
• Add dependency-reduced-pom to .gitignore by @alexbaden in #385
• Enable IN test, remove string-related queries. by @kurapov-peter in #384
• Add conda flow for intel gpus by @kurapov-peter in #360
• [NaN handling] Support NaN proper processing. by @Devjiu in #337
• Add the LLVM MulOverFlow, extend IntelGPUtest by @lmontigny in #383
• Add shared memory pointers address space by @kurapov-peter in #386
• Pass opt/debug options to clang compiler on windows by @gshimansky in #398
• Fixed int64_t constants on windows by @gshimansky in #401
• Add ResultSetRegistry storage [3/N] by @ienkovich in #365
• Improve issue bucketing by @kurapov-peter in #376
• Temporary disable string queries for L0 by @akroviakov in #388
• Fixed generation of left shift on windows by @gshimansky in #408
• Disable inline of CodegenTraits::get to prevent cross-DLL access to descCallingConvToLLVM by @gshimansky in #405
• Revert a docker part of PR 366 by @leshikus in #412
• Fixed call to gmtime_s because error code is ignored on Linux by @gshimansky in #407
• Disable optimizing out function bodies on windows by @gshimansky in #402
• Fixed -DENABLE_JIT_DEBUG=on mode by @gshimansky in #399
• Fixed hash code of empty query plan on windows by @gshimansky in #400
• Fixed inline functions declaration on windows by @gshimansky in #406
• Prevent zero-sized copies to/from GPU by @kurapov-peter in #414
• Unified build flow by @kurapov-peter in #362
• Force deps resolving on link. by @ienkovich in #382
• Fixed python build on windows after ResultSet changes by @gshimansky in #404
• Linear regression backup by @bagrorg in #389
• [CI] Enable Python style check. by @Devjiu in #347
• [CI] Rename jobs by @Devjiu in #387
• Fixed definition of atomics on windows by @gshimansky in #397
• Resolve a circular dependency between CostModel and QueryEngine libraries by @kurapov-peter in #419
• Remove unused header include by @alexbaden in #420
• [Cython Build] Pass all compile definitions by @Devjiu in #410
• Use a single arena per RowSetMemoryOwner. by @ienkovich in #421
• Support varlen arrays in Arrow converter. by @ienkovich in #411
• Added dllimport/dllexport where necessary by @gshimansky in #250
• Support fixed length arrays in Arrow converter. by @ienkovich in #426
• Updated conda env for windows development by @gshimansky in #403
• [CI][Naming] Shorten jobs name by @Devjiu in #422
• Add ResultSetRegistry storage [4/N] by @ienkovich in #413
• Add a missed flag by @leshikus in #380
• Added missing libraries to linking by @gshimansky in #431
• Create Calcite singleton using a dedicated worker thread by @alexbaden in #416
• Add ResultSetRegistry storage [5/N] by @ienkovich in #428
• Initial windows CI that should build release from sources by @leshikus in #433
• Remove Unused String Dictionary Storage Code by @alexbaden in #441
• Refactor Node::replaceInput by @ienkovich in #444
• Remove const from input nodes. by @ienkovich in #450
• Support POWER in query builder. by @ienkovich in #451
• [CI] Naming by @Devjiu in #448
• Pin to Apache Arrow v11 by @alexbaden in #449
• Fixed "cmake install" on windows by @gshimansky in #445
• Add ResultSetRegistry storage [6/N] by @ienkovich in #442
• Fix warnings. by @ienkovich in #456
• [Null schema] Support emty data table. by @Devjiu in #446
• Add ResultSetRegistry storage [7/7] by @ienkovich in #459
• Use functions to map column indexes to IDs. by @ienkovich in #460
• Fixed finding root path when HDK is installed in conda env and is called from python by @gshimansky in #463
• Remove persistence level from table info. by @ienkovich in #462
• Added call to os.add_dll_directory base on JAVA_HOME value by @gshimansky in #464
• Don't use streaming-top-n in intel gpu case by @kurapov-peter in #455
• Support STDDEV aggregate through query builder. by @ienkovich in #467
• Add API to fetch a row from pyhdk.ExecutionResult. by @ienkovich in #466
• Fix rows order in parallel fetch for projections and sorted results. by @ienkovich in #443
• Cost model integration by @bagrorg in #423
• Support shape property for scan nodes. by @ienkovich in #465
• Support CORR aggregate through query builder. by @ienkovich in #470
• Add head and tail API for execution results by @ienkovich in #469
• Improve detection of cases for large NDV estimator. by @ienkovich in #475
• Improve transient string dict error msg logging by @alexbaden in #476
• Provide 'shared' implementations for count and sum functions by @kurapov-peter in #479
• Ignore logging options in ConfigBuilder. by @ienkovich in #474
• Don't override CMAKE_INSTALL_PREFIX if it is already set for us by @gshimansky in #473
• Add Intel GPU build instructions by @kurapov-peter in #485
• Temporary fallback for hash joins on L0 by @akroviakov in #477
• Disable most compiler warnings on windows by @gshimansky in #484
• Add support for DISTINCT clause by @kurapov-peter in #480
• Don't run sort on gpu in L0 case by @kurapov-peter in #483
• Lazily materialize string dictionaries by @alexbaden in #481
• Add a clang build by @leshikus in #471

Full Changelog: v0.5.0...v0.6.0