diff --git a/.github/workflows/build.yaml b/.github/workflows/build.yaml
index 666d8844a80..e27361ab263 100644
--- a/.github/workflows/build.yaml
+++ b/.github/workflows/build.yaml
@@ -22,13 +22,13 @@ on:
default: nightly
concurrency:
- group: ${{ github.workflow }}-${{ github.ref }}
+ group: ${{ github.workflow }}-${{ github.ref }}-${{ github.event_name }}
cancel-in-progress: true
jobs:
cpp-build:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
branch: ${{ inputs.branch }}
@@ -37,7 +37,7 @@ jobs:
python-build:
needs: [cpp-build]
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-build.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
branch: ${{ inputs.branch }}
@@ -46,7 +46,7 @@ jobs:
upload-conda:
needs: [cpp-build, python-build]
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
branch: ${{ inputs.branch }}
@@ -57,7 +57,7 @@ jobs:
if: github.ref_type == 'branch'
needs: python-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
arch: "amd64"
branch: ${{ inputs.branch }}
@@ -69,9 +69,10 @@ jobs:
sha: ${{ inputs.sha }}
wheel-build-cudf:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
+ build-2_28-wheels: "true"
branch: ${{ inputs.branch }}
sha: ${{ inputs.sha }}
date: ${{ inputs.date }}
@@ -79,7 +80,7 @@ jobs:
wheel-publish-cudf:
needs: wheel-build-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
branch: ${{ inputs.branch }}
@@ -89,7 +90,7 @@ jobs:
wheel-build-dask-cudf:
needs: wheel-publish-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64" and .PY_VER == "3.10" and (.CUDA_VER == "11.8.0" or .CUDA_VER == "12.0.1")))
build_type: ${{ inputs.build_type || 'branch' }}
@@ -100,7 +101,7 @@ jobs:
wheel-publish-dask-cudf:
needs: wheel-build-dask-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-publish.yaml@branch-23.12
with:
build_type: ${{ inputs.build_type || 'branch' }}
branch: ${{ inputs.branch }}
diff --git a/.github/workflows/labeler.yml b/.github/workflows/labeler.yml
index 23956a02fbd..31e78f82a62 100644
--- a/.github/workflows/labeler.yml
+++ b/.github/workflows/labeler.yml
@@ -6,6 +6,6 @@ jobs:
triage:
runs-on: ubuntu-latest
steps:
- - uses: actions/labeler@main
+ - uses: actions/labeler@v4
with:
repo-token: "${{ secrets.GITHUB_TOKEN }}"
diff --git a/.github/workflows/pr.yaml b/.github/workflows/pr.yaml
index abf5fcf2f33..40cf0dcd2c1 100644
--- a/.github/workflows/pr.yaml
+++ b/.github/workflows/pr.yaml
@@ -30,34 +30,34 @@ jobs:
#- pandas-tests-diff
#- pandas-tests-diff-comment
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/pr-builder.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/pr-builder.yaml@branch-23.12
checks:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/checks.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/checks.yaml@branch-23.12
with:
enable_check_generated_files: false
conda-cpp-build:
needs: checks
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-23.12
with:
build_type: pull-request
conda-cpp-tests:
needs: conda-cpp-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-tests.yaml@branch-23.12
with:
build_type: pull-request
conda-python-build:
needs: conda-cpp-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-build.yaml@branch-23.12
with:
build_type: pull-request
conda-python-cudf-tests:
needs: conda-python-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.12
with:
build_type: pull-request
test_script: "ci/test_python_cudf.sh"
@@ -65,14 +65,14 @@ jobs:
# Tests for dask_cudf, custreamz, cudf_kafka are separated for CI parallelism
needs: conda-python-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.12
with:
build_type: pull-request
test_script: "ci/test_python_other.sh"
conda-java-tests:
needs: conda-cpp-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: pull-request
node_type: "gpu-v100-latest-1"
@@ -82,7 +82,7 @@ jobs:
conda-notebook-tests:
needs: conda-python-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: pull-request
node_type: "gpu-v100-latest-1"
@@ -92,7 +92,7 @@ jobs:
docs-build:
needs: conda-python-build
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: pull-request
node_type: "gpu-v100-latest-1"
@@ -102,21 +102,22 @@ jobs:
wheel-build-cudf:
needs: checks
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.12
with:
build_type: pull-request
+ build-2_28-wheels: "true"
script: "ci/build_wheel_cudf.sh"
wheel-tests-cudf:
needs: wheel-build-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
build_type: pull-request
script: ci/test_wheel_cudf.sh
wheel-build-dask-cudf:
needs: wheel-tests-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-build.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64" and .PY_VER == "3.10" and (.CUDA_VER == "11.8.0" or .CUDA_VER == "12.0.1")))
build_type: pull-request
@@ -124,7 +125,7 @@ jobs:
wheel-tests-dask-cudf:
needs: wheel-build-dask-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64" and .PY_VER == "3.10" and (.CUDA_VER == "11.8.0" or .CUDA_VER == "12.0.1")))
build_type: pull-request
@@ -132,7 +133,7 @@ jobs:
unit-tests-cudf-pandas:
needs: wheel-build-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64" and .PY_VER == "3.10" and (.CUDA_VER == "11.8.0" or .CUDA_VER == "12.0.1")))
build_type: pull-request
@@ -141,7 +142,7 @@ jobs:
# run the Pandas unit tests using PR branch
needs: wheel-build-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64")) | max_by(.CUDA_VER) | [.]
build_type: pull-request
diff --git a/.github/workflows/test.yaml b/.github/workflows/test.yaml
index 49a9c73d026..0d4401160e1 100644
--- a/.github/workflows/test.yaml
+++ b/.github/workflows/test.yaml
@@ -16,7 +16,7 @@ on:
jobs:
conda-cpp-tests:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-tests.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -24,7 +24,7 @@ jobs:
sha: ${{ inputs.sha }}
conda-cpp-memcheck-tests:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -36,7 +36,7 @@ jobs:
run_script: "ci/test_cpp_memcheck.sh"
conda-python-cudf-tests:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -46,7 +46,7 @@ jobs:
conda-python-other-tests:
# Tests for dask_cudf, custreamz, cudf_kafka are separated for CI parallelism
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/conda-python-tests.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -55,7 +55,7 @@ jobs:
test_script: "ci/test_python_other.sh"
conda-java-tests:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -67,7 +67,7 @@ jobs:
run_script: "ci/test_java.sh"
conda-notebook-tests:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -79,7 +79,7 @@ jobs:
run_script: "ci/test_notebooks.sh"
wheel-tests-cudf:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -88,7 +88,7 @@ jobs:
script: ci/test_wheel_cudf.sh
wheel-tests-dask-cudf:
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64" and .PY_VER == "3.10" and (.CUDA_VER == "11.8.0" or .CUDA_VER == "12.0.1")))
build_type: nightly
@@ -97,9 +97,8 @@ jobs:
sha: ${{ inputs.sha }}
script: ci/test_wheel_dask_cudf.sh
unit-tests-cudf-pandas:
- needs: wheel-build-cudf
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
build_type: nightly
branch: ${{ inputs.branch }}
@@ -109,7 +108,7 @@ jobs:
pandas-tests:
# run the Pandas unit tests
secrets: inherit
- uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.10
+ uses: rapidsai/shared-workflows/.github/workflows/wheels-test.yaml@branch-23.12
with:
matrix_filter: map(select(.ARCH == "amd64")) | max_by(.CUDA_VER) | [.]
build_type: nightly
diff --git a/CHANGELOG.md b/CHANGELOG.md
index ecd547ab5b3..3cb6caa25ee 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,3 +1,140 @@
+# cuDF 23.12.00 (6 Dec 2023)
+
+## 🚨 Breaking Changes
+
+- Raise error in `reindex` when `index` is not unique ([#14400](https://github.com/rapidsai/cudf/pull/14400)) [@galipremsagar](https://github.com/galipremsagar)
+- Expose stream parameter to get_json_object API ([#14297](https://github.com/rapidsai/cudf/pull/14297)) [@davidwendt](https://github.com/davidwendt)
+- Refactor cudf_kafka to use skbuild ([#14292](https://github.com/rapidsai/cudf/pull/14292)) [@jdye64](https://github.com/jdye64)
+- Expose stream parameter in public strings convert APIs ([#14255](https://github.com/rapidsai/cudf/pull/14255)) [@davidwendt](https://github.com/davidwendt)
+- Upgrade to nvCOMP 3.0.4 ([#13815](https://github.com/rapidsai/cudf/pull/13815)) [@vuule](https://github.com/vuule)
+
+## 🐛 Bug Fixes
+
+- Update actions/labeler to v4 ([#14562](https://github.com/rapidsai/cudf/pull/14562)) [@raydouglass](https://github.com/raydouglass)
+- Fix data corruption when skipping rows ([#14557](https://github.com/rapidsai/cudf/pull/14557)) [@etseidl](https://github.com/etseidl)
+- Fix function name typo in `cudf.pandas` profiler ([#14514](https://github.com/rapidsai/cudf/pull/14514)) [@galipremsagar](https://github.com/galipremsagar)
+- Fix intermediate type checking in expression parsing ([#14445](https://github.com/rapidsai/cudf/pull/14445)) [@vyasr](https://github.com/vyasr)
+- Forward merge `branch-23.10` into `branch-23.12` ([#14435](https://github.com/rapidsai/cudf/pull/14435)) [@raydouglass](https://github.com/raydouglass)
+- Remove needs: wheel-build-cudf. ([#14427](https://github.com/rapidsai/cudf/pull/14427)) [@bdice](https://github.com/bdice)
+- Fix dask dependency in custreamz ([#14420](https://github.com/rapidsai/cudf/pull/14420)) [@vyasr](https://github.com/vyasr)
+- Ensure nvbench initializes nvml context when built statically ([#14411](https://github.com/rapidsai/cudf/pull/14411)) [@robertmaynard](https://github.com/robertmaynard)
+- Support java AST String literal with desired encoding ([#14402](https://github.com/rapidsai/cudf/pull/14402)) [@winningsix](https://github.com/winningsix)
+- Raise error in `reindex` when `index` is not unique ([#14400](https://github.com/rapidsai/cudf/pull/14400)) [@galipremsagar](https://github.com/galipremsagar)
+- Always build nvbench statically so we don't need to package it ([#14399](https://github.com/rapidsai/cudf/pull/14399)) [@robertmaynard](https://github.com/robertmaynard)
+- Fix token-count logic in nvtext::tokenize_with_vocabulary ([#14393](https://github.com/rapidsai/cudf/pull/14393)) [@davidwendt](https://github.com/davidwendt)
+- Fix as_column(pd.Timestamp/Timedelta, length=) not respecting length ([#14390](https://github.com/rapidsai/cudf/pull/14390)) [@mroeschke](https://github.com/mroeschke)
+- cudf.pandas: cuDF subpath checking in module `__getattr__` ([#14388](https://github.com/rapidsai/cudf/pull/14388)) [@shwina](https://github.com/shwina)
+- Fix and disable encoding for nanosecond statistics in ORC writer ([#14367](https://github.com/rapidsai/cudf/pull/14367)) [@vuule](https://github.com/vuule)
+- Add the new manylinux builds to the build job ([#14351](https://github.com/rapidsai/cudf/pull/14351)) [@vyasr](https://github.com/vyasr)
+- cudf jit parser now supports .pragma instructions with quotes ([#14348](https://github.com/rapidsai/cudf/pull/14348)) [@robertmaynard](https://github.com/robertmaynard)
+- Fix overflow check in `cudf::merge` ([#14345](https://github.com/rapidsai/cudf/pull/14345)) [@divyegala](https://github.com/divyegala)
+- Add cramjam ([#14344](https://github.com/rapidsai/cudf/pull/14344)) [@vyasr](https://github.com/vyasr)
+- Enable `dask_cudf/io` pytests in CI ([#14338](https://github.com/rapidsai/cudf/pull/14338)) [@galipremsagar](https://github.com/galipremsagar)
+- Temporarily avoid the current build of pydata-sphinx-theme ([#14332](https://github.com/rapidsai/cudf/pull/14332)) [@vyasr](https://github.com/vyasr)
+- Fix host buffer access from device function in the Parquet reader ([#14328](https://github.com/rapidsai/cudf/pull/14328)) [@vuule](https://github.com/vuule)
+- Run IO tests for Dask-cuDF ([#14327](https://github.com/rapidsai/cudf/pull/14327)) [@rjzamora](https://github.com/rjzamora)
+- Fix logical type issues in the Parquet writer ([#14322](https://github.com/rapidsai/cudf/pull/14322)) [@vuule](https://github.com/vuule)
+- Remove aws-sdk-pinning and revert to arrow 12.0.1 ([#14319](https://github.com/rapidsai/cudf/pull/14319)) [@vyasr](https://github.com/vyasr)
+- test is_valid before reading column data ([#14318](https://github.com/rapidsai/cudf/pull/14318)) [@etseidl](https://github.com/etseidl)
+- Fix gtest validity setting for TextTokenizeTest.Vocabulary ([#14312](https://github.com/rapidsai/cudf/pull/14312)) [@davidwendt](https://github.com/davidwendt)
+- Fixes stack context for json lines format that recovers from invalid JSON lines ([#14309](https://github.com/rapidsai/cudf/pull/14309)) [@elstehle](https://github.com/elstehle)
+- Downgrade to Arrow 12.0.0 for aws-sdk-cpp and fix cudf_kafka builds for new CI containers ([#14296](https://github.com/rapidsai/cudf/pull/14296)) [@vyasr](https://github.com/vyasr)
+- fixing thread index overflow issue ([#14290](https://github.com/rapidsai/cudf/pull/14290)) [@hyperbolic2346](https://github.com/hyperbolic2346)
+- Fix memset error in nvtext::edit_distance_matrix ([#14283](https://github.com/rapidsai/cudf/pull/14283)) [@davidwendt](https://github.com/davidwendt)
+- Changes JSON reader's recovery option's behaviour to ignore all characters after a valid JSON record ([#14279](https://github.com/rapidsai/cudf/pull/14279)) [@elstehle](https://github.com/elstehle)
+- Handle empty string correctly in Parquet statistics ([#14257](https://github.com/rapidsai/cudf/pull/14257)) [@etseidl](https://github.com/etseidl)
+- Fixes behaviour for incomplete lines when `recover_with_nulls` is enabled ([#14252](https://github.com/rapidsai/cudf/pull/14252)) [@elstehle](https://github.com/elstehle)
+- cudf::detail::pinned_allocator doesn't throw from `deallocate` ([#14251](https://github.com/rapidsai/cudf/pull/14251)) [@robertmaynard](https://github.com/robertmaynard)
+- Fix strings replace for adjacent, identical multi-byte UTF-8 character targets ([#14235](https://github.com/rapidsai/cudf/pull/14235)) [@davidwendt](https://github.com/davidwendt)
+- Fix the precision when converting a decimal128 column to an arrow array ([#14230](https://github.com/rapidsai/cudf/pull/14230)) [@jihoonson](https://github.com/jihoonson)
+- Fixing parquet list of struct interpretation ([#13715](https://github.com/rapidsai/cudf/pull/13715)) [@hyperbolic2346](https://github.com/hyperbolic2346)
+
+## 📖 Documentation
+
+- Fix io reference in docs. ([#14452](https://github.com/rapidsai/cudf/pull/14452)) [@bdice](https://github.com/bdice)
+- Update README ([#14374](https://github.com/rapidsai/cudf/pull/14374)) [@shwina](https://github.com/shwina)
+- Example code for blog on new row comparators ([#13795](https://github.com/rapidsai/cudf/pull/13795)) [@divyegala](https://github.com/divyegala)
+
+## 🚀 New Features
+
+- Expose streams in public unary APIs ([#14342](https://github.com/rapidsai/cudf/pull/14342)) [@vyasr](https://github.com/vyasr)
+- Add python tests for Parquet DELTA_BINARY_PACKED encoder ([#14316](https://github.com/rapidsai/cudf/pull/14316)) [@etseidl](https://github.com/etseidl)
+- Update rapids-cmake functions to non-deprecated signatures ([#14265](https://github.com/rapidsai/cudf/pull/14265)) [@robertmaynard](https://github.com/robertmaynard)
+- Expose streams in public null mask APIs ([#14263](https://github.com/rapidsai/cudf/pull/14263)) [@vyasr](https://github.com/vyasr)
+- Expose streams in binaryop APIs ([#14187](https://github.com/rapidsai/cudf/pull/14187)) [@vyasr](https://github.com/vyasr)
+- Add pylibcudf.Scalar that interoperates with Arrow scalars ([#14133](https://github.com/rapidsai/cudf/pull/14133)) [@vyasr](https://github.com/vyasr)
+- Add decoder for DELTA_BYTE_ARRAY to Parquet reader ([#14101](https://github.com/rapidsai/cudf/pull/14101)) [@etseidl](https://github.com/etseidl)
+- Add DELTA_BINARY_PACKED encoder for Parquet writer ([#14100](https://github.com/rapidsai/cudf/pull/14100)) [@etseidl](https://github.com/etseidl)
+- Add BytePairEncoder class to cuDF ([#13891](https://github.com/rapidsai/cudf/pull/13891)) [@davidwendt](https://github.com/davidwendt)
+- Upgrade to nvCOMP 3.0.4 ([#13815](https://github.com/rapidsai/cudf/pull/13815)) [@vuule](https://github.com/vuule)
+- Use `pynvjitlink` for CUDA 12+ MVC ([#13650](https://github.com/rapidsai/cudf/pull/13650)) [@brandon-b-miller](https://github.com/brandon-b-miller)
+
+## 🛠️ Improvements
+
+- Build concurrency for nightly and merge triggers ([#14441](https://github.com/rapidsai/cudf/pull/14441)) [@bdice](https://github.com/bdice)
+- Cleanup remaining usages of dask dependencies ([#14407](https://github.com/rapidsai/cudf/pull/14407)) [@galipremsagar](https://github.com/galipremsagar)
+- Update to Arrow 14.0.1. ([#14387](https://github.com/rapidsai/cudf/pull/14387)) [@bdice](https://github.com/bdice)
+- Remove Cython libcpp wrappers ([#14382](https://github.com/rapidsai/cudf/pull/14382)) [@vyasr](https://github.com/vyasr)
+- Forward-merge branch-23.10 to branch-23.12 ([#14372](https://github.com/rapidsai/cudf/pull/14372)) [@bdice](https://github.com/bdice)
+- Upgrade to arrow 14 ([#14371](https://github.com/rapidsai/cudf/pull/14371)) [@galipremsagar](https://github.com/galipremsagar)
+- Fix a pytest typo in `test_kurt_skew_error` ([#14368](https://github.com/rapidsai/cudf/pull/14368)) [@galipremsagar](https://github.com/galipremsagar)
+- Use new rapids-dask-dependency metapackage for managing dask versions ([#14364](https://github.com/rapidsai/cudf/pull/14364)) [@vyasr](https://github.com/vyasr)
+- Change `nullable()` to `has_nulls()` in `cudf::detail::gather` ([#14363](https://github.com/rapidsai/cudf/pull/14363)) [@divyegala](https://github.com/divyegala)
+- Split up scan_inclusive.cu to improve its compile time ([#14358](https://github.com/rapidsai/cudf/pull/14358)) [@davidwendt](https://github.com/davidwendt)
+- Implement user_datasource_wrapper is_empty() and is_device_read_preferred(). ([#14357](https://github.com/rapidsai/cudf/pull/14357)) [@tpn](https://github.com/tpn)
+- Added streams to CSV reader and writer api ([#14340](https://github.com/rapidsai/cudf/pull/14340)) [@shrshi](https://github.com/shrshi)
+- Upgrade wheels to use arrow 13 ([#14339](https://github.com/rapidsai/cudf/pull/14339)) [@vyasr](https://github.com/vyasr)
+- Rework nvtext::byte_pair_encoding API ([#14337](https://github.com/rapidsai/cudf/pull/14337)) [@davidwendt](https://github.com/davidwendt)
+- Improve performance of nvtext::tokenize_with_vocabulary for long strings ([#14336](https://github.com/rapidsai/cudf/pull/14336)) [@davidwendt](https://github.com/davidwendt)
+- Upgrade `arrow` to `13` ([#14330](https://github.com/rapidsai/cudf/pull/14330)) [@galipremsagar](https://github.com/galipremsagar)
+- Expose stream parameter in public nvtext replace APIs ([#14329](https://github.com/rapidsai/cudf/pull/14329)) [@davidwendt](https://github.com/davidwendt)
+- Drop `pyorc` dependency and use `pandas`/`pyarrow` instead ([#14323](https://github.com/rapidsai/cudf/pull/14323)) [@galipremsagar](https://github.com/galipremsagar)
+- Avoid `pyarrow.fs` import for local storage ([#14321](https://github.com/rapidsai/cudf/pull/14321)) [@rjzamora](https://github.com/rjzamora)
+- Unpin `dask` and `distributed` for `23.12` development ([#14320](https://github.com/rapidsai/cudf/pull/14320)) [@galipremsagar](https://github.com/galipremsagar)
+- Expose stream parameter in public nvtext tokenize APIs ([#14317](https://github.com/rapidsai/cudf/pull/14317)) [@davidwendt](https://github.com/davidwendt)
+- Added streams to JSON reader and writer api ([#14313](https://github.com/rapidsai/cudf/pull/14313)) [@shrshi](https://github.com/shrshi)
+- Minor improvements in `source_info` ([#14308](https://github.com/rapidsai/cudf/pull/14308)) [@vuule](https://github.com/vuule)
+- Forward-merge branch-23.10 to branch-23.12 ([#14307](https://github.com/rapidsai/cudf/pull/14307)) [@bdice](https://github.com/bdice)
+- Add stream parameter to Set Operations (Public List APIs) ([#14305](https://github.com/rapidsai/cudf/pull/14305)) [@SurajAralihalli](https://github.com/SurajAralihalli)
+- Expose stream parameter to get_json_object API ([#14297](https://github.com/rapidsai/cudf/pull/14297)) [@davidwendt](https://github.com/davidwendt)
+- Sort dictionary data alphabetically in the ORC writer ([#14295](https://github.com/rapidsai/cudf/pull/14295)) [@vuule](https://github.com/vuule)
+- Expose stream parameter in public strings filter APIs ([#14293](https://github.com/rapidsai/cudf/pull/14293)) [@davidwendt](https://github.com/davidwendt)
+- Refactor cudf_kafka to use skbuild ([#14292](https://github.com/rapidsai/cudf/pull/14292)) [@jdye64](https://github.com/jdye64)
+- Update `shared-action-workflows` references ([#14289](https://github.com/rapidsai/cudf/pull/14289)) [@AyodeAwe](https://github.com/AyodeAwe)
+- Register ``partd`` encode dispatch in ``dask_cudf`` ([#14287](https://github.com/rapidsai/cudf/pull/14287)) [@rjzamora](https://github.com/rjzamora)
+- Update versioning strategy ([#14285](https://github.com/rapidsai/cudf/pull/14285)) [@vyasr](https://github.com/vyasr)
+- Move and rename byte-pair-encoding source files ([#14284](https://github.com/rapidsai/cudf/pull/14284)) [@davidwendt](https://github.com/davidwendt)
+- Expose stream parameter in public strings combine APIs ([#14281](https://github.com/rapidsai/cudf/pull/14281)) [@davidwendt](https://github.com/davidwendt)
+- Expose stream parameter in public strings contains APIs ([#14280](https://github.com/rapidsai/cudf/pull/14280)) [@davidwendt](https://github.com/davidwendt)
+- Add stream parameter to List Sort and Filter APIs ([#14272](https://github.com/rapidsai/cudf/pull/14272)) [@SurajAralihalli](https://github.com/SurajAralihalli)
+- Use branch-23.12 workflows. ([#14271](https://github.com/rapidsai/cudf/pull/14271)) [@bdice](https://github.com/bdice)
+- Refactor LogicalType for Parquet ([#14264](https://github.com/rapidsai/cudf/pull/14264)) [@etseidl](https://github.com/etseidl)
+- Centralize chunked reading code in the parquet reader to reader_impl_chunking.cu ([#14262](https://github.com/rapidsai/cudf/pull/14262)) [@nvdbaranec](https://github.com/nvdbaranec)
+- Expose stream parameter in public strings replace APIs ([#14261](https://github.com/rapidsai/cudf/pull/14261)) [@davidwendt](https://github.com/davidwendt)
+- Expose stream parameter in public strings APIs ([#14260](https://github.com/rapidsai/cudf/pull/14260)) [@davidwendt](https://github.com/davidwendt)
+- Cleanup of namespaces in parquet code. ([#14259](https://github.com/rapidsai/cudf/pull/14259)) [@nvdbaranec](https://github.com/nvdbaranec)
+- Make parquet schema index type consistent ([#14256](https://github.com/rapidsai/cudf/pull/14256)) [@hyperbolic2346](https://github.com/hyperbolic2346)
+- Expose stream parameter in public strings convert APIs ([#14255](https://github.com/rapidsai/cudf/pull/14255)) [@davidwendt](https://github.com/davidwendt)
+- Add in java bindings for DataSource ([#14254](https://github.com/rapidsai/cudf/pull/14254)) [@revans2](https://github.com/revans2)
+- Reimplement `cudf::merge` for nested types without using comparators ([#14250](https://github.com/rapidsai/cudf/pull/14250)) [@divyegala](https://github.com/divyegala)
+- Add stream parameter to List Manipulation and Operations APIs ([#14248](https://github.com/rapidsai/cudf/pull/14248)) [@SurajAralihalli](https://github.com/SurajAralihalli)
+- Expose stream parameter in public strings split/partition APIs ([#14247](https://github.com/rapidsai/cudf/pull/14247)) [@davidwendt](https://github.com/davidwendt)
+- Improve `contains_column` by invoking `contains_table` ([#14238](https://github.com/rapidsai/cudf/pull/14238)) [@PointKernel](https://github.com/PointKernel)
+- Detect and report errors in Parquet header parsing ([#14237](https://github.com/rapidsai/cudf/pull/14237)) [@etseidl](https://github.com/etseidl)
+- Normalizing offsets iterator ([#14234](https://github.com/rapidsai/cudf/pull/14234)) [@davidwendt](https://github.com/davidwendt)
+- Forward merge `23.10` into `23.12` ([#14231](https://github.com/rapidsai/cudf/pull/14231)) [@galipremsagar](https://github.com/galipremsagar)
+- Return error if BOOL8 column-type is used with integers-to-hex ([#14208](https://github.com/rapidsai/cudf/pull/14208)) [@davidwendt](https://github.com/davidwendt)
+- Enable indexalator for device code ([#14206](https://github.com/rapidsai/cudf/pull/14206)) [@davidwendt](https://github.com/davidwendt)
+- Marginally reduce memory footprint of joins ([#14197](https://github.com/rapidsai/cudf/pull/14197)) [@wence-](https://github.com/wence-)
+- Add nvtx annotations to spilling-based data movement ([#14196](https://github.com/rapidsai/cudf/pull/14196)) [@wence-](https://github.com/wence-)
+- Optimize ORC writer for decimal columns ([#14190](https://github.com/rapidsai/cudf/pull/14190)) [@vuule](https://github.com/vuule)
+- Remove the use of volatile in ORC ([#14175](https://github.com/rapidsai/cudf/pull/14175)) [@vuule](https://github.com/vuule)
+- Add `bytes_per_second` to distinct_count of stream_compaction nvbench. ([#14172](https://github.com/rapidsai/cudf/pull/14172)) [@Blonck](https://github.com/Blonck)
+- Add `bytes_per_second` to transpose benchmark ([#14170](https://github.com/rapidsai/cudf/pull/14170)) [@Blonck](https://github.com/Blonck)
+- cuDF: Build CUDA 12.0 ARM conda packages. ([#14112](https://github.com/rapidsai/cudf/pull/14112)) [@bdice](https://github.com/bdice)
+- Add `bytes_per_second` to shift benchmark ([#13950](https://github.com/rapidsai/cudf/pull/13950)) [@Blonck](https://github.com/Blonck)
+- Extract `debug_utilities.hpp/cu` from `column_utilities.hpp/cu` ([#13720](https://github.com/rapidsai/cudf/pull/13720)) [@ttnghia](https://github.com/ttnghia)
+
# cuDF 23.10.00 (11 Oct 2023)
## 🚨 Breaking Changes
diff --git a/README.md b/README.md
index 64c980d0cb3..677cfc89d52 100644
--- a/README.md
+++ b/README.md
@@ -1,57 +1,62 @@
# 
cuDF - GPU DataFrames
gather(table_view const& source_table,
mr));
}
- auto const nullable = bounds_policy == out_of_bounds_policy::NULLIFY ||
- std::any_of(source_table.begin(), source_table.end(), [](auto const& col) {
- return col.nullable();
- });
- if (nullable) {
- auto const op = bounds_policy == out_of_bounds_policy::NULLIFY ? gather_bitmask_op::NULLIFY
- : gather_bitmask_op::DONT_CHECK;
- gather_bitmask(source_table, gather_map_begin, destination_columns, op, stream, mr);
+ auto needs_new_bitmask = bounds_policy == out_of_bounds_policy::NULLIFY ||
+ cudf::has_nested_nullable_columns(source_table);
+ if (needs_new_bitmask) {
+ needs_new_bitmask = needs_new_bitmask || cudf::has_nested_nulls(source_table);
+ if (needs_new_bitmask) {
+ auto const op = bounds_policy == out_of_bounds_policy::NULLIFY
+ ? gather_bitmask_op::NULLIFY
+ : gather_bitmask_op::DONT_CHECK;
+ gather_bitmask(source_table, gather_map_begin, destination_columns, op, stream, mr);
+ } else {
+ for (size_type i = 0; i < source_table.num_columns(); ++i) {
+ set_all_valid_null_masks(source_table.column(i), *destination_columns[i], stream, mr);
+ }
+ }
}
return std::make_unique(std::move(destination_columns));
diff --git a/cpp/include/cudf/detail/indexalator.cuh b/cpp/include/cudf/detail/indexalator.cuh
index 6532dae3695..4d261c54b29 100644
--- a/cpp/include/cudf/detail/indexalator.cuh
+++ b/cpp/include/cudf/detail/indexalator.cuh
@@ -56,10 +56,69 @@ namespace detail {
* auto result = thrust::find(thrust::device, begin, end, size_type{12} );
* @endcode
*/
-using input_indexalator = input_normalator;
+struct input_indexalator : base_normalator {
+ friend struct base_normalator; // for CRTP
+
+ using reference = cudf::size_type const; // this keeps STL and thrust happy
+
+ input_indexalator() = default;
+ input_indexalator(input_indexalator const&) = default;
+ input_indexalator(input_indexalator&&) = default;
+ input_indexalator& operator=(input_indexalator const&) = default;
+ input_indexalator& operator=(input_indexalator&&) = default;
+
+ /**
+ * @brief Indirection operator returns the value at the current iterator position
+ */
+ __device__ inline cudf::size_type operator*() const { return operator[](0); }
+
+ /**
+ * @brief Dispatch functor for resolving a Integer value from any integer type
+ */
+ struct normalize_type {
+ template ())>
+ __device__ cudf::size_type operator()(void const* tp)
+ {
+ return static_cast(*static_cast(tp));
+ }
+ template ())>
+ __device__ cudf::size_type operator()(void const*)
+ {
+ CUDF_UNREACHABLE("only integral types are supported");
+ }
+ };
+
+ /**
+ * @brief Array subscript operator returns a value at the input
+ * `idx` position as a `Integer` value.
+ */
+ __device__ inline cudf::size_type operator[](size_type idx) const
+ {
+ void const* tp = p_ + (idx * this->width_);
+ return type_dispatcher(this->dtype_, normalize_type{}, tp);
+ }
+
+ /**
+ * @brief Create an input index normalizing iterator
+ *
+ * Use the indexalator_factory to create an iterator instance.
+ *
+ * @param data Pointer to an integer array in device memory.
+ * @param dtype Type of data in data
+ * @param offset Applied to the data pointer per size of the type
+ */
+ CUDF_HOST_DEVICE input_indexalator(void const* data, data_type dtype, cudf::size_type offset = 0)
+ : base_normalator(dtype), p_{static_cast(data)}
+ {
+ p_ += offset * this->width_;
+ }
+
+ protected:
+ char const* p_; /// pointer to the integer data in device memory
+};
/**
- * @brief The index normalizing output iterator.
+ * @brief The index normalizing output iterator
*
* This is an iterator that can be used for index types (integers) without
* requiring a type-specific instance. It can be used for any iterator
@@ -82,7 +141,75 @@ using input_indexalator = input_normalator;
* thrust::less());
* @endcode
*/
-using output_indexalator = output_normalator;
+struct output_indexalator : base_normalator {
+ friend struct base_normalator; // for CRTP
+
+ using reference = output_indexalator const&; // required for output iterators
+
+ output_indexalator() = default;
+ output_indexalator(output_indexalator const&) = default;
+ output_indexalator(output_indexalator&&) = default;
+ output_indexalator& operator=(output_indexalator const&) = default;
+ output_indexalator& operator=(output_indexalator&&) = default;
+
+ /**
+ * @brief Indirection operator returns this iterator instance in order
+ * to capture the `operator=(Integer)` calls.
+ */
+ __device__ inline reference operator*() const { return *this; }
+
+ /**
+ * @brief Array subscript operator returns an iterator instance at the specified `idx` position.
+ *
+ * This allows capturing the subsequent `operator=(Integer)` call in this class.
+ */
+ __device__ inline output_indexalator const operator[](size_type idx) const
+ {
+ output_indexalator tmp{*this};
+ tmp.p_ += (idx * this->width_);
+ return tmp;
+ }
+
+ /**
+ * @brief Dispatch functor for setting the index value from a size_type value.
+ */
+ struct normalize_type {
+ template ())>
+ __device__ void operator()(void* tp, cudf::size_type const value)
+ {
+ (*static_cast(tp)) = static_cast(value);
+ }
+ template ())>
+ __device__ void operator()(void*, cudf::size_type const)
+ {
+ CUDF_UNREACHABLE("only index types are supported");
+ }
+ };
+
+ /**
+ * @brief Assign an Integer value to the current iterator position
+ */
+ __device__ inline reference operator=(cudf::size_type const value) const
+ {
+ void* tp = p_;
+ type_dispatcher(this->dtype_, normalize_type{}, tp, value);
+ return *this;
+ }
+
+ /**
+ * @brief Create an output normalizing iterator
+ *
+ * @param data Pointer to an integer array in device memory.
+ * @param dtype Type of data in data
+ */
+ CUDF_HOST_DEVICE output_indexalator(void* data, data_type dtype)
+ : base_normalator(dtype), p_{static_cast(data)}
+ {
+ }
+
+ protected:
+ char* p_; /// pointer to the integer data in device memory
+};
/**
* @brief Use this class to create an indexalator instance.
@@ -92,14 +219,12 @@ struct indexalator_factory {
* @brief A type_dispatcher functor to create an input iterator from an indices column.
*/
struct input_indexalator_fn {
- template ()>* = nullptr>
+ template ())>
input_indexalator operator()(column_view const& indices)
{
return input_indexalator(indices.data(), indices.type());
}
- template ()>* = nullptr>
+ template ())>
input_indexalator operator()(Args&&... args)
{
CUDF_FAIL("indices must be an index type");
@@ -110,16 +235,14 @@ struct indexalator_factory {
* @brief Use this class to create an indexalator to a scalar index.
*/
struct input_indexalator_scalar_fn {
- template ()>* = nullptr>
+ template ())>
input_indexalator operator()(scalar const& index)
{
// note: using static_cast const&>(index) creates a copy
auto const scalar_impl = static_cast const*>(&index);
return input_indexalator(scalar_impl->data(), index.type());
}
- template ()>* = nullptr>
+ template ())>
input_indexalator operator()(Args&&... args)
{
CUDF_FAIL("scalar must be an index type");
@@ -130,14 +253,12 @@ struct indexalator_factory {
* @brief A type_dispatcher functor to create an output iterator from an indices column.
*/
struct output_indexalator_fn {
- template ()>* = nullptr>
+ template ())>
output_indexalator operator()(mutable_column_view const& indices)
{
return output_indexalator(indices.data(), indices.type());
}
- template ()>* = nullptr>
+ template ())>
output_indexalator operator()(Args&&... args)
{
CUDF_FAIL("indices must be an index type");
diff --git a/cpp/include/cudf/detail/interop.hpp b/cpp/include/cudf/detail/interop.hpp
index 44024333239..8124471982d 100644
--- a/cpp/include/cudf/detail/interop.hpp
+++ b/cpp/include/cudf/detail/interop.hpp
@@ -194,5 +194,18 @@ std::unique_ptr from_arrow(arrow::Table const& input_table,
std::unique_ptr from_arrow(arrow::Scalar const& input,
rmm::cuda_stream_view stream,
rmm::mr::device_memory_resource* mr);
+
+/**
+ * @brief Return a maximum precision for a given type.
+ *
+ * @tparam T the type to get the maximum precision for
+ */
+template
+constexpr std::size_t max_precision()
+{
+ auto constexpr num_bits = sizeof(T) * 8;
+ return std::floor(num_bits * std::log(2) / std::log(10));
+}
+
} // namespace detail
} // namespace cudf
diff --git a/cpp/include/cudf/detail/merge.cuh b/cpp/include/cudf/detail/merge.cuh
deleted file mode 100644
index e8e9b080a92..00000000000
--- a/cpp/include/cudf/detail/merge.cuh
+++ /dev/null
@@ -1,166 +0,0 @@
-/*
- * Copyright (c) 2018-2022, NVIDIA CORPORATION.
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- * http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#pragma once
-
-#include
-#include
-
-#include
-#include
-#include
-
-namespace cudf {
-namespace detail {
-/**
- * @brief Source table identifier to copy data from.
- */
-enum class side : bool { LEFT, RIGHT };
-
-/**
- * @brief Tagged index type: `thrust::get<0>` indicates left/right side,
- * `thrust::get<1>` indicates the row index
- */
-using index_type = thrust::pair;
-
-/**
- * @brief Vector of `index_type` values.
- */
-using index_vector = rmm::device_uvector;
-
-/**
- * @brief tagged_element_relational_comparator uses element_relational_comparator to provide
- * "tagged-index" comparison logic.
- *
- * Special treatment is necessary in several thrust algorithms (e.g., merge()) where
- * the index affinity to the side is not guaranteed; i.e., the algorithms rely on
- * binary functors (predicates) where the operands may transparently switch sides.
- *
- * For example,
- * thrust::merge(left_container,
- * right_container,
- * predicate(lhs, rhs){...});
- * can create 4 different use-cases, inside predicate(...):
- *
- * 1. lhs refers to the left container; rhs to the right container;
- * 2. vice-versa;
- * 3. both lhs and rhs actually refer to the left container;
- * 4. both lhs and rhs actually refer to the right container;
- *
- * Because of that, one cannot rely on the predicate having *fixed* references to the containers.
- * Each invocation may land in a different situation (among the 4 above) than any other invocation.
- * Also, one cannot just manipulate lhs, rhs (indices) alone; because, if predicate always applies
- * one index to one container and the other index to the other container,
- * switching the indices alone won't suffice in the cases (3) or (4),
- * where the also the containers must be changed (to just one instead of two)
- * independently of indices;
- *
- * As a result, a special comparison logic is necessary whereby the index is "tagged" with side
- * information and consequently comparator functors (predicates) must operate on these tagged
- * indices rather than on raw indices.
- */
-template
-struct tagged_element_relational_comparator {
- __host__ __device__ tagged_element_relational_comparator(column_device_view lhs,
- column_device_view rhs,
- null_order null_precedence)
- : lhs{lhs}, rhs{rhs}, null_precedence{null_precedence}
- {
- }
-
- [[nodiscard]] __device__ weak_ordering compare(index_type lhs_tagged_index,
- index_type rhs_tagged_index) const noexcept
- {
- auto const [l_side, l_indx] = lhs_tagged_index;
- auto const [r_side, r_indx] = rhs_tagged_index;
-
- column_device_view const* ptr_left_dview{l_side == side::LEFT ? &lhs : &rhs};
- column_device_view const* ptr_right_dview{r_side == side::LEFT ? &lhs : &rhs};
-
- auto erl_comparator = element_relational_comparator(
- nullate::DYNAMIC{has_nulls}, *ptr_left_dview, *ptr_right_dview, null_precedence);
-
- return cudf::type_dispatcher(lhs.type(), erl_comparator, l_indx, r_indx);
- }
-
- private:
- column_device_view lhs;
- column_device_view rhs;
- null_order null_precedence;
-};
-
-/**
- * @brief The equivalent of `row_lexicographic_comparator` for tagged indices.
- */
-template
-struct row_lexicographic_tagged_comparator {
- row_lexicographic_tagged_comparator(table_device_view lhs,
- table_device_view rhs,
- order const* column_order = nullptr,
- null_order const* null_precedence = nullptr)
- : _lhs{lhs}, _rhs{rhs}, _column_order{column_order}, _null_precedence{null_precedence}
- {
- // Add check for types to be the same.
- CUDF_EXPECTS(_lhs.num_columns() == _rhs.num_columns(), "Mismatched number of columns.");
- }
-
- __device__ bool operator()(index_type lhs_tagged_index,
- index_type rhs_tagged_index) const noexcept
- {
- for (size_type i = 0; i < _lhs.num_columns(); ++i) {
- bool ascending = (_column_order == nullptr) or (_column_order[i] == order::ASCENDING);
-
- null_order null_precedence =
- _null_precedence == nullptr ? null_order::BEFORE : _null_precedence[i];
-
- auto comparator = tagged_element_relational_comparator{
- _lhs.column(i), _rhs.column(i), null_precedence};
-
- weak_ordering state = comparator.compare(lhs_tagged_index, rhs_tagged_index);
-
- if (state == weak_ordering::EQUIVALENT) { continue; }
-
- return state == (ascending ? weak_ordering::LESS : weak_ordering::GREATER);
- }
- return false;
- }
-
- private:
- table_device_view _lhs;
- table_device_view _rhs;
- null_order const* _null_precedence{};
- order const* _column_order{};
-};
-
-/**
- * @copydoc std::unique_ptr merge(
- * std::vector const& tables_to_merge,
- * std::vector const& key_cols,
- * std::vector const& column_order,
- * std::vector const& null_precedence,
- * rmm::mr::device_memory_resource* mr)
- *
- * @param stream CUDA stream used for device memory operations and kernel launches
- */
-std::unique_ptr merge(std::vector const& tables_to_merge,
- std::vector const& key_cols,
- std::vector const& column_order,
- std::vector const& null_precedence,
- rmm::cuda_stream_view stream,
- rmm::mr::device_memory_resource* mr);
-
-} // namespace detail
-} // namespace cudf
diff --git a/cpp/include/cudf/detail/merge.hpp b/cpp/include/cudf/detail/merge.hpp
new file mode 100644
index 00000000000..2167a484214
--- /dev/null
+++ b/cpp/include/cudf/detail/merge.hpp
@@ -0,0 +1,60 @@
+/*
+ * Copyright (c) 2018-2023, NVIDIA CORPORATION.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include
+
+#include
+
+namespace cudf {
+namespace detail {
+
+/**
+ * @brief Source table identifier to copy data from.
+ */
+enum class side : bool { LEFT, RIGHT };
+
+/**
+ * @brief Tagged index type: `thrust::get<0>` indicates left/right side,
+ * `thrust::get<1>` indicates the row index
+ */
+using index_type = thrust::pair;
+
+/**
+ * @brief Vector of `index_type` values.
+ */
+using index_vector = rmm::device_uvector;
+
+/**
+ * @copydoc std::unique_ptr merge(
+ * std::vector const& tables_to_merge,
+ * std::vector const& key_cols,
+ * std::vector const& column_order,
+ * std::vector const& null_precedence,
+ * rmm::mr::device_memory_resource* mr)
+ *
+ * @param stream CUDA stream used for device memory operations and kernel launches
+ */
+std::unique_ptr merge(std::vector const& tables_to_merge,
+ std::vector const& key_cols,
+ std::vector const& column_order,
+ std::vector const& null_precedence,
+ rmm::cuda_stream_view stream,
+ rmm::mr::device_memory_resource* mr);
+
+} // namespace detail
+} // namespace cudf
diff --git a/cpp/include/cudf/detail/normalizing_iterator.cuh b/cpp/include/cudf/detail/normalizing_iterator.cuh
index 51b3133f84f..8f90afc3e57 100644
--- a/cpp/include/cudf/detail/normalizing_iterator.cuh
+++ b/cpp/include/cudf/detail/normalizing_iterator.cuh
@@ -33,8 +33,8 @@ namespace detail {
* @tparam Integer The type the iterator normalizes to
*/
template
-struct base_normalator {
- static_assert(std::is_integral_v);
+struct alignas(16) base_normalator {
+ static_assert(cudf::is_index_type());
using difference_type = std::ptrdiff_t;
using value_type = Integer;
using pointer = Integer*;
@@ -202,165 +202,43 @@ struct base_normalator {
return static_cast(*this).p_ >= rhs.p_;
}
- protected:
- /**
- * @brief Constructor assigns width and type member variables for base class.
- */
- explicit base_normalator(data_type dtype) : width_(size_of(dtype)), dtype_(dtype) {}
-
- int width_; /// integer type width = 1,2,4, or 8
- data_type dtype_; /// for type-dispatcher calls
-};
-
-/**
- * @brief The integer normalizing input iterator
- *
- * This is an iterator that can be used for index types (integers) without
- * requiring a type-specific instance. It can be used for any iterator
- * interface for reading an array of integer values of type
- * int8, int16, int32, int64, uint8, uint16, uint32, or uint64.
- * Reading specific elements always return a type of `Integer`
- *
- * @tparam Integer Type returned by all read functions
- */
-template
-struct input_normalator : base_normalator, Integer> {
- friend struct base_normalator, Integer>; // for CRTP
-
- using reference = Integer const; // this keeps STL and thrust happy
-
- input_normalator() = default;
- input_normalator(input_normalator const&) = default;
- input_normalator(input_normalator&&) = default;
- input_normalator& operator=(input_normalator const&) = default;
- input_normalator& operator=(input_normalator&&) = default;
-
- /**
- * @brief Indirection operator returns the value at the current iterator position
- */
- __device__ inline Integer operator*() const { return operator[](0); }
-
- /**
- * @brief Dispatch functor for resolving a Integer value from any integer type
- */
- struct normalize_type {
- template >* = nullptr>
- __device__ Integer operator()(void const* tp)
- {
- return static_cast(*static_cast(tp));
- }
- template >* = nullptr>
- __device__ Integer operator()(void const*)
+ private:
+ struct integer_sizeof_fn {
+ template ())>
+ CUDF_HOST_DEVICE constexpr std::size_t operator()() const
{
+#ifndef __CUDA_ARCH__
+ CUDF_FAIL("only integral types are supported");
+#else
CUDF_UNREACHABLE("only integral types are supported");
+#endif
}
- };
-
- /**
- * @brief Array subscript operator returns a value at the input
- * `idx` position as a `Integer` value.
- */
- __device__ inline Integer operator[](size_type idx) const
- {
- void const* tp = p_ + (idx * this->width_);
- return type_dispatcher(this->dtype_, normalize_type{}, tp);
- }
-
- /**
- * @brief Create an input index normalizing iterator.
- *
- * Use the indexalator_factory to create an iterator instance.
- *
- * @param data Pointer to an integer array in device memory.
- * @param data_type Type of data in data
- */
- input_normalator(void const* data, data_type dtype)
- : base_normalator, Integer>(dtype), p_{static_cast(data)}
- {
- }
-
- char const* p_; /// pointer to the integer data in device memory
-};
-
-/**
- * @brief The integer normalizing output iterator
- *
- * This is an iterator that can be used for index types (integers) without
- * requiring a type-specific instance. It can be used for any iterator
- * interface for writing an array of integer values of type
- * int8, int16, int32, int64, uint8, uint16, uint32, or uint64.
- * Setting specific elements always accept the `Integer` type values.
- *
- * @tparam Integer The type used for all write functions
- */
-template
-struct output_normalator : base_normalator, Integer> {
- friend struct base_normalator, Integer>; // for CRTP
-
- using reference = output_normalator const&; // required for output iterators
-
- output_normalator() = default;
- output_normalator(output_normalator const&) = default;
- output_normalator(output_normalator&&) = default;
- output_normalator& operator=(output_normalator const&) = default;
- output_normalator& operator=(output_normalator&&) = default;
-
- /**
- * @brief Indirection operator returns this iterator instance in order
- * to capture the `operator=(Integer)` calls.
- */
- __device__ inline output_normalator const& operator*() const { return *this; }
-
- /**
- * @brief Array subscript operator returns an iterator instance at the specified `idx` position.
- *
- * This allows capturing the subsequent `operator=(Integer)` call in this class.
- */
- __device__ inline output_normalator const operator[](size_type idx) const
- {
- output_normalator tmp{*this};
- tmp.p_ += (idx * this->width_);
- return tmp;
- }
-
- /**
- * @brief Dispatch functor for setting the index value from a size_type value.
- */
- struct normalize_type {
- template >* = nullptr>
- __device__ void operator()(void* tp, Integer const value)
- {
- (*static_cast(tp)) = static_cast(value);
- }
- template >* = nullptr>
- __device__ void operator()(void*, Integer const)
+ template ())>
+ CUDF_HOST_DEVICE constexpr std::size_t operator()() const noexcept
{
- CUDF_UNREACHABLE("only index types are supported");
+ return sizeof(T);
}
};
+ protected:
/**
- * @brief Assign an Integer value to the current iterator position
+ * @brief Constructor assigns width and type member variables for base class.
*/
- __device__ inline output_normalator const& operator=(Integer const value) const
+ explicit CUDF_HOST_DEVICE base_normalator(data_type dtype) : dtype_(dtype)
{
- void* tp = p_;
- type_dispatcher(this->dtype_, normalize_type{}, tp, value);
- return *this;
+ width_ = static_cast(type_dispatcher(dtype, integer_sizeof_fn{}));
}
/**
- * @brief Create an output normalizing iterator
- *
- * @param data Pointer to an integer array in device memory.
- * @param data_type Type of data in data
+ * @brief Constructor assigns width and type member variables for base class.
*/
- output_normalator(void* data, data_type dtype)
- : base_normalator, Integer>(dtype), p_{static_cast(data)}
+ explicit CUDF_HOST_DEVICE base_normalator(data_type dtype, int32_t width)
+ : width_(width), dtype_(dtype)
{
}
- char* p_; /// pointer to the integer data in device memory
+ int32_t width_; /// integer type width = 1,2,4, or 8
+ data_type dtype_; /// for type-dispatcher calls
};
} // namespace detail
diff --git a/cpp/include/cudf/detail/null_mask.hpp b/cpp/include/cudf/detail/null_mask.hpp
index 8c10bbe416f..74e2ccd2ea1 100644
--- a/cpp/include/cudf/detail/null_mask.hpp
+++ b/cpp/include/cudf/detail/null_mask.hpp
@@ -15,6 +15,7 @@
*/
#pragma once
+#include
#include
#include
#include
@@ -259,6 +260,22 @@ cudf::size_type inplace_bitmask_and(device_span dest_mask,
size_type mask_size_bits,
rmm::cuda_stream_view stream);
+/**
+ * @brief Recursively set valid null masks for all children.
+ *
+ * This function applies all valid null masks to the output column if input column satisfies
+ * `nullable() == true` condition
+ *
+ * @param input input column to check for nullability
+ * @param output output column to mirror nullability of input
+ * @param stream CUDA stream used for device memory operations and kernel launches
+ * @param mr Device memory resource used to allocate the returned column's device memory
+ */
+void set_all_valid_null_masks(column_view const& input,
+ column& output,
+ rmm::cuda_stream_view stream,
+ rmm::mr::device_memory_resource* mr);
+
} // namespace detail
} // namespace cudf
diff --git a/cpp/include/cudf/detail/offsets_iterator.cuh b/cpp/include/cudf/detail/offsets_iterator.cuh
new file mode 100644
index 00000000000..3eb77b32353
--- /dev/null
+++ b/cpp/include/cudf/detail/offsets_iterator.cuh
@@ -0,0 +1,165 @@
+/*
+ * Copyright (c) 2023, NVIDIA CORPORATION.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include
+#include
+
+namespace cudf {
+namespace detail {
+
+/**
+ * @brief The offsets normalizing input iterator
+ *
+ * This is an iterator that can be used for offsets where the underlying
+ * type may be int32_t or int64_t.
+ *
+ * Use the offsetalator_factory to create an appropriate input iterator
+ * from an offsets column_view.
+ */
+struct input_offsetalator : base_normalator {
+ friend struct base_normalator; // for CRTP
+
+ using reference = int64_t const; // this keeps STL and thrust happy
+
+ input_offsetalator() = default;
+ input_offsetalator(input_offsetalator const&) = default;
+ input_offsetalator(input_offsetalator&&) = default;
+ input_offsetalator& operator=(input_offsetalator const&) = default;
+ input_offsetalator& operator=(input_offsetalator&&) = default;
+
+ /**
+ * @brief Indirection operator returns the value at the current iterator position
+ */
+ __device__ inline int64_t operator*() const { return operator[](0); }
+
+ /**
+ * @brief Array subscript operator returns a value at the input
+ * `idx` position as a int64_t value.
+ */
+ __device__ inline int64_t operator[](size_type idx) const
+ {
+ void const* tp = p_ + (idx * this->width_);
+ return this->width_ == sizeof(int32_t) ? static_cast(*static_cast(tp))
+ : *static_cast(tp);
+ }
+
+ /**
+ * @brief Create an input index normalizing iterator.
+ *
+ * Use the indexalator_factory to create an iterator instance.
+ *
+ * @param data Pointer to an integer array in device memory.
+ * @param dtype Type of data in data
+ */
+ CUDF_HOST_DEVICE input_offsetalator(void const* data, data_type dtype)
+ : base_normalator(
+ dtype, dtype.id() == type_id::INT32 ? sizeof(int32_t) : sizeof(int64_t)),
+ p_{static_cast(data)}
+ {
+#ifndef __CUDA_ARCH__
+ CUDF_EXPECTS(dtype.id() == type_id::INT32 || dtype.id() == type_id::INT64,
+ "Unexpected offsets type");
+#else
+ cudf_assert((dtype.id() == type_id::INT32 || dtype.id() == type_id::INT64) &&
+ "Unexpected offsets type");
+#endif
+ }
+
+ protected:
+ char const* p_; /// pointer to the integer data in device memory
+};
+
+/**
+ * @brief The offsets normalizing output iterator
+ *
+ * This is an iterator that can be used for storing offsets values
+ * where the underlying type may be either int32_t or int64_t.
+ *
+ * Use the offsetalator_factory to create an appropriate output iterator
+ * from a mutable_column_view.
+ *
+ */
+struct output_offsetalator : base_normalator {
+ friend struct base_normalator; // for CRTP
+
+ using reference = output_offsetalator const&; // required for output iterators
+
+ output_offsetalator() = default;
+ output_offsetalator(output_offsetalator const&) = default;
+ output_offsetalator(output_offsetalator&&) = default;
+ output_offsetalator& operator=(output_offsetalator const&) = default;
+ output_offsetalator& operator=(output_offsetalator&&) = default;
+
+ /**
+ * @brief Indirection operator returns this iterator instance in order
+ * to capture the `operator=(int64)` calls.
+ */
+ __device__ inline output_offsetalator const& operator*() const { return *this; }
+
+ /**
+ * @brief Array subscript operator returns an iterator instance at the specified `idx` position.
+ *
+ * This allows capturing the subsequent `operator=(int64)` call in this class.
+ */
+ __device__ inline output_offsetalator const operator[](size_type idx) const
+ {
+ output_offsetalator tmp{*this};
+ tmp.p_ += (idx * this->width_);
+ return tmp;
+ }
+
+ /**
+ * @brief Assign an offset value to the current iterator position
+ */
+ __device__ inline output_offsetalator const& operator=(int64_t const value) const
+ {
+ void* tp = p_;
+ if (this->width_ == sizeof(int32_t)) {
+ (*static_cast(tp)) = static_cast(value);
+ } else {
+ (*static_cast(tp)) = value;
+ }
+ return *this;
+ }
+
+ /**
+ * @brief Create an output offsets iterator
+ *
+ * @param data Pointer to an integer array in device memory.
+ * @param dtype Type of data in data
+ */
+ CUDF_HOST_DEVICE output_offsetalator(void* data, data_type dtype)
+ : base_normalator(
+ dtype, dtype.id() == type_id::INT32 ? sizeof(int32_t) : sizeof(int64_t)),
+ p_{static_cast(data)}
+ {
+#ifndef __CUDA_ARCH__
+ CUDF_EXPECTS(dtype.id() == type_id::INT32 || dtype.id() == type_id::INT64,
+ "Unexpected offsets type");
+#else
+ cudf_assert((dtype.id() == type_id::INT32 || dtype.id() == type_id::INT64) &&
+ "Unexpected offsets type");
+#endif
+ }
+
+ protected:
+ char* p_; /// pointer to the integer data in device memory
+};
+
+} // namespace detail
+} // namespace cudf
diff --git a/cpp/include/cudf/detail/offsets_iterator_factory.cuh b/cpp/include/cudf/detail/offsets_iterator_factory.cuh
new file mode 100644
index 00000000000..5b4c6b825d2
--- /dev/null
+++ b/cpp/include/cudf/detail/offsets_iterator_factory.cuh
@@ -0,0 +1,47 @@
+/*
+ * Copyright (c) 2023, NVIDIA CORPORATION.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include
+#include
+
+namespace cudf {
+namespace detail {
+
+/**
+ * @brief Use this class to create an offsetalator instance.
+ */
+struct offsetalator_factory {
+ /**
+ * @brief Create an input offsetalator instance from an offsets column
+ */
+ static input_offsetalator make_input_iterator(column_view const& offsets)
+ {
+ return input_offsetalator(offsets.head(), offsets.type());
+ }
+
+ /**
+ * @brief Create an output offsetalator instance from an offsets column
+ */
+ static output_offsetalator make_output_iterator(mutable_column_view const& offsets)
+ {
+ return output_offsetalator(offsets.head(), offsets.type());
+ }
+};
+
+} // namespace detail
+} // namespace cudf
diff --git a/cpp/include/cudf/detail/unary.hpp b/cpp/include/cudf/detail/unary.hpp
index 3fbdf4a5a8f..12f864de572 100644
--- a/cpp/include/cudf/detail/unary.hpp
+++ b/cpp/include/cudf/detail/unary.hpp
@@ -64,8 +64,6 @@ std::unique_ptr true_if(InputIterator begin,
/**
* @copydoc cudf::unary_operation
- *
- * @param stream CUDA stream used for device memory operations and kernel launches.
*/
std::unique_ptr unary_operation(cudf::column_view const& input,
cudf::unary_operator op,
@@ -74,8 +72,6 @@ std::unique_ptr unary_operation(cudf::column_view const& input,
/**
* @copydoc cudf::is_valid
- *
- * @param stream CUDA stream used for device memory operations and kernel launches.
*/
std::unique_ptr is_valid(cudf::column_view const& input,
rmm::cuda_stream_view stream,
@@ -83,8 +79,6 @@ std::unique_ptr is_valid(cudf::column_view const& input,
/**
* @copydoc cudf::cast
- *
- * @param stream CUDA stream used for device memory operations and kernel launches.
*/
std::unique_ptr cast(column_view const& input,
data_type type,
@@ -93,8 +87,6 @@ std::unique_ptr cast(column_view const& input,
/**
* @copydoc cudf::is_nan
- *
- * @param[in] stream CUDA stream used for device memory operations and kernel launches.
*/
std::unique_ptr