Stop using versioneer to manage versions (#12741)

This PR replaces usage of versioneer with hard-coded version numbers in setup.py and __init__.py. Since cudf needs to manage versions across a wide range of file types (CMake, C++, Sphinx and doxygen docs, etc), versioneer cannot be relied on as a single source of truth and therefore does not allow us to single-source our versioning to the Git repo as is intended. Additionally, since the primary means of installing cudf is via conda packages (or now, pip packages), information from the package manager tends to be far more informative than the version strings for troubleshooting and debugging purposes. Conversely, the nonstandard version strings that it produces tend to be problematic for other tools, which at best will ignore such versions but at worst will simply fail.

This PR also replaces usage of an environment variable to set the package name for wheels in setup.py, instead moving the renaming logic into the same sed script used to update package versions. This change makes setup.py essentially static, paving the way for migration to pyproject.toml.

Authors:
  - Vyas Ramasubramani (https://github.com/vyasr)

Approvers:
  - AJ Schmidt (https://github.com/ajschmidt8)
  - Lawrence Mitchell (https://github.com/wence-)

URL: #12741

.github/workflows/build.yaml

@@ -64,6 +64,7 @@ jobs:
       package-name: cudf
       package-dir: python/cudf
       skbuild-configure-options: "-DCUDF_BUILD_WHEELS=ON -DDETECT_CONDA_ENV=OFF"
+      uses-setup-env-vars: false
   wheel-publish-cudf:
     needs: wheel-build-cudf
     secrets: inherit
@@ -85,6 +86,7 @@ jobs:
       date: ${{ inputs.date }}
       package-name: dask_cudf
       package-dir: python/dask_cudf
+      uses-setup-env-vars: false
   wheel-publish-dask-cudf:
     needs: wheel-build-dask-cudf
     secrets: inherit

.github/workflows/pr.yaml

@@ -91,6 +91,7 @@ jobs:
       package-name: cudf
       package-dir: python/cudf
       skbuild-configure-options: "-DCUDF_BUILD_WHEELS=ON -DDETECT_CONDA_ENV=OFF"
+      uses-setup-env-vars: false
   wheel-tests-cudf:
     needs: wheel-build-cudf
     secrets: inherit
@@ -112,6 +113,7 @@ jobs:
       package-name: dask_cudf
       package-dir: python/dask_cudf
       before-wheel: "RAPIDS_PY_WHEEL_NAME=cudf_cu11 rapids-download-wheels-from-s3 ./local-cudf && pip install --no-deps ./local-cudf/cudf*.whl"
+      uses-setup-env-vars: false
   wheel-tests-dask-cudf:
     needs: wheel-build-dask-cudf
     secrets: inherit

.pre-commit-config.yaml

@@ -144,8 +144,7 @@ repos:
                 exclude: |
                   (?x)^(
                     .*test.*|
-                    ^CHANGELOG.md$|
-                    ^.*versioneer.py$
+                    ^CHANGELOG.md$
                   )
 
 default_language_version:

ci/checks/copyright.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2019-2022, NVIDIA CORPORATION.
+# Copyright (c) 2019-2023, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -31,7 +31,6 @@
 ]
 ExemptFiles = [
     re.compile(r"cpp/include/cudf_test/cxxopts.hpp"),
-    re.compile(r"versioneer[.]py"),
 ]
 
 # this will break starting at year 10000, which is probably OK :)

ci/release/apply_wheel_modifications.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+# Copyright (c) 2023, NVIDIA CORPORATION.
+#
+# Usage: bash apply_wheel_modifications.sh <new_version> <cuda_suffix>
+
+VERSION=${1}
+CUDA_SUFFIX=${2}
+
+# __init__.py versions
+sed -i "s/__version__ = .*/__version__ = \"${VERSION}\"/g" python/cudf/cudf/__init__.py
+sed -i "s/__version__ = .*/__version__ = \"${VERSION}\"/g" python/dask_cudf/dask_cudf/__init__.py
+sed -i "s/__version__ = .*/__version__ = \"${VERSION}\"/g" python/cudf_kafka/cudf_kafka/__init__.py
+sed -i "s/__version__ = .*/__version__ = \"${VERSION}\"/g" python/custreamz/custreamz/__init__.py
+
+# setup.py versions
+sed -i "s/version=.*,/version=\"${VERSION}\",/g" python/cudf/setup.py
+sed -i "s/version=.*,/version=\"${VERSION}\",/g" python/dask_cudf/setup.py
+sed -i "s/version=.*,/version=\"${VERSION}\",/g" python/cudf_kafka/setup.py
+sed -i "s/version=.*,/version=\"${VERSION}\",/g" python/custreamz/setup.py
+
+# cudf setup.py cuda suffixes
+sed -i "s/name=\"cudf\"/name=\"cudf${CUDA_SUFFIX}\"/g" python/cudf/setup.py
+sed -i "s/rmm/rmm${CUDA_SUFFIX}/g" python/cudf/setup.py
+sed -i "s/ptxcompiler/ptxcompiler${CUDA_SUFFIX}/g" python/cudf/setup.py
+sed -i "s/cubinlinker/cubinlinker${CUDA_SUFFIX}/g" python/cudf/setup.py
+
+# cudf pyproject.toml cuda suffixes
+sed -i "s/rmm/rmm${CUDA_SUFFIX}/g" python/cudf/pyproject.toml
+
+# dask_cudf setup.py cuda suffixes
+sed -i "s/name=\"dask-cudf\"/name=\"dask-cudf${CUDA_SUFFIX}\"/g" python/dask_cudf/setup.py
+# Need to provide the == to avoid modifying the URL
+sed -i "s/\"cudf==/\"cudf${CUDA_SUFFIX}==/g" python/dask_cudf/setup.py

ci/release/update-version.sh

@@ -34,19 +34,27 @@ function sed_runner() {
 # cpp update
 sed_runner 's/'"VERSION ${CURRENT_SHORT_TAG}.*"'/'"VERSION ${NEXT_FULL_TAG}"'/g' cpp/CMakeLists.txt
 
-# cpp stream testing update
-sed_runner 's/'"VERSION ${CURRENT_SHORT_TAG}.*"'/'"VERSION ${NEXT_FULL_TAG}"'/g' cpp/tests/utilities/identify_stream_usage/CMakeLists.txt
-
-# Python update
+# Python CMakeLists updates
 sed_runner 's/'"cudf_version .*)"'/'"cudf_version ${NEXT_FULL_TAG})"'/g' python/cudf/CMakeLists.txt
 
-
 # cpp libcudf_kafka update
 sed_runner 's/'"VERSION ${CURRENT_SHORT_TAG}.*"'/'"VERSION ${NEXT_FULL_TAG}"'/g' cpp/libcudf_kafka/CMakeLists.txt
 
 # cpp cudf_jni update
 sed_runner 's/'"VERSION ${CURRENT_SHORT_TAG}.*"'/'"VERSION ${NEXT_FULL_TAG}"'/g' java/src/main/native/CMakeLists.txt
 
+# Python __init__.py updates
+sed_runner "s/__version__ = .*/__version__ = \"${NEXT_FULL_TAG}\"/g" python/cudf/cudf/__init__.py
+sed_runner "s/__version__ = .*/__version__ = \"${NEXT_FULL_TAG}\"/g" python/dask_cudf/dask_cudf/__init__.py
+sed_runner "s/__version__ = .*/__version__ = \"${NEXT_FULL_TAG}\"/g" python/cudf_kafka/cudf_kafka/__init__.py
+sed_runner "s/__version__ = .*/__version__ = \"${NEXT_FULL_TAG}\"/g" python/custreamz/custreamz/__init__.py
+
+# Python setup.py updates
+sed_runner "s/version=.*,/version=\"${NEXT_FULL_TAG}\",/g" python/cudf/setup.py
+sed_runner "s/version=.*,/version=\"${NEXT_FULL_TAG}\",/g" python/dask_cudf/setup.py
+sed_runner "s/version=.*,/version=\"${NEXT_FULL_TAG}\",/g" python/cudf_kafka/setup.py
+sed_runner "s/version=.*,/version=\"${NEXT_FULL_TAG}\",/g" python/custreamz/setup.py
+
 # rapids-cmake version
 sed_runner 's/'"branch-.*\/RAPIDS.cmake"'/'"branch-${NEXT_SHORT_TAG}\/RAPIDS.cmake"'/g' fetch_rapids.cmake
 
@@ -81,9 +89,12 @@ sed_runner "s/CUDF_TAG branch-${CURRENT_SHORT_TAG}/CUDF_TAG branch-${NEXT_SHORT_
 # Need to distutils-normalize the original version
 NEXT_SHORT_TAG_PEP440=$(python -c "from setuptools.extern import packaging; print(packaging.version.Version('${NEXT_SHORT_TAG}'))")
 
-# Wheel builds install intra-RAPIDS dependencies from same release
-sed_runner "s/rmm{cuda_suffix}.*\",/rmm{cuda_suffix}==${NEXT_SHORT_TAG_PEP440}.*\",/g" python/cudf/setup.py
-sed_runner "s/cudf{cuda_suffix}==.*\",/cudf{cuda_suffix}==${NEXT_SHORT_TAG_PEP440}.*\",/g" python/dask_cudf/setup.py
+# Dependency versions in setup.py
+sed_runner "s/rmm==.*\",/rmm==${NEXT_SHORT_TAG_PEP440}.*\",/g" python/cudf/setup.py
+sed_runner "s/cudf==.*\",/cudf==${NEXT_SHORT_TAG_PEP440}.*\",/g" python/dask_cudf/setup.py
+
+# Dependency versions in pyproject.toml
+sed_runner "s/rmm==.*\",/rmm==${NEXT_SHORT_TAG_PEP440}.*\",/g" python/pyproject.toml
 
 for FILE in .github/workflows/*.yaml; do
   sed_runner "/shared-action-workflows/ s/@.*/@branch-${NEXT_SHORT_TAG}/g" "${FILE}"

python/cudf/cudf/__init__.py

@@ -10,7 +10,6 @@
 import rmm
 
 from cudf import api, core, datasets, testing
-from cudf._version import get_versions
 from cudf.api.extensions import (
     register_dataframe_accessor,
     register_index_accessor,
@@ -112,8 +111,7 @@
 rmm.register_reinitialize_hook(clear_cache)
 
 
-__version__ = get_versions()["version"]
-del get_versions
+__version__ = "23.04.00"
 
 __all__ = [
     "BaseIndex",