fix(backends): make string concat-with-null behavior consistent across backends #18990

Workflow file for this run

.github/workflows/ibis-backends.yml at 0ab9d81

	name: Backends

	on:
	push:
	# Skip the backend suite if all changes are docs
	paths-ignore:
	- "docs/**"
	- "*/.md"
	- "*/.qmd"
	- "codecov.yml"
	- ".envrc"
	- ".codespellrc"
	branches:
	- main
	- "*.x.x"
	- the-epic-split
	pull_request:
	# Skip the backend suite if all changes are docs
	paths-ignore:
	- "docs/**"
	- "*/.md"
	- "*/.qmd"
	- "codecov.yml"
	- ".envrc"
	- ".codespellrc"
	branches:
	- main
	- "*.x.x"
	- the-epic-split
	merge_group:

	permissions:
	# this allows extractions/setup-just to list releases for `just` at a higher
	# rate limit while restricting GITHUB_TOKEN permissions elsewhere
	contents: read

	concurrency:
	group: ${{ github.repository }}-${{ github.head_ref \|\| github.sha }}-${{ github.workflow }}
	cancel-in-progress: true

	env:
	FORCE_COLOR: "1"
	ODBCSYSINI: "${{ github.workspace }}/ci/odbc"
	HYPOTHESIS_PROFILE: "ci"

	jobs:
	test_backends:
	name: ${{ matrix.backend.title }} ${{ matrix.os }} python-${{ matrix.python-version }}
	runs-on: ${{ matrix.os }}
	strategy:
	fail-fast: false
	matrix:
	os:
	- ubuntu-latest
	- windows-latest
	python-version:
	- "3.9"
	- "3.11"
	backend:
	- name: duckdb
	title: DuckDB
	extras:
	- duckdb
	- deltalake
	- geospatial
	- examples
	- decompiler
	additional_deps:
	- torch
	- name: clickhouse
	title: ClickHouse
	services:
	- clickhouse
	extras:
	- clickhouse
	- examples
	- name: dask
	title: Dask
	extras:
	- dask
	- name: pandas
	title: Pandas
	extras:
	- pandas
	- name: sqlite
	title: SQLite
	extras:
	- sqlite
	- name: datafusion
	title: Datafusion
	extras:
	- datafusion
	- name: polars
	title: Polars
	extras:
	- polars
	- deltalake
	- name: mysql
	title: MySQL
	services:
	- mysql
	extras:
	- mysql
	- geospatial
	sys-deps:
	- libgeos-dev
	- name: postgres
	title: PostgreSQL
	extras:
	- postgres
	- geospatial
	services:
	- postgres
	sys-deps:
	- libgeos-dev
	- name: postgres
	title: PostgreSQL + Torch
	extras:
	- postgres
	- geospatial
	additional_deps:
	- torch
	services:
	- postgres
	sys-deps:
	- libgeos-dev
	- name: risingwave
	title: Risingwave
	services:
	- risingwave
	extras:
	- risingwave
	- name: impala
	title: Impala
	serial: true
	extras:
	- impala
	services:
	- impala
	- kudu
	sys-deps:
	- cmake
	- ninja-build
	- name: mssql
	title: MS SQL Server
	extras:
	- mssql
	services:
	- mssql
	sys-deps:
	- freetds-dev
	- unixodbc-dev
	- tdsodbc
	- name: trino
	title: Trino
	extras:
	- trino
	services:
	- trino
	- name: druid
	title: Druid
	extras:
	- druid
	services:
	- druid
	- name: exasol
	title: Exasol
	serial: true
	extras:
	- exasol
	services:
	- exasol
	- name: oracle
	title: Oracle
	serial: true
	extras:
	- oracle
	services:
	- oracle
	# - name: flink
	# title: Flink
	# serial: true
	# extras:
	# - flink
	# additional_deps:
	# - apache-flink
	# - pytest-split
	# services:
	# - flink
	# - name: risingwave
	# title: Risingwave
	# services:
	# - risingwave
	# extras:
	# - risingwave
	exclude:
	- os: windows-latest
	backend:
	name: mysql
	title: MySQL
	extras:
	- mysql
	- geospatial
	services:
	- mysql
	sys-deps:
	- libgeos-dev
	- os: windows-latest
	backend:
	name: clickhouse
	title: ClickHouse
	extras:
	- clickhouse
	- examples
	services:
	- clickhouse
	- os: windows-latest
	backend:
	name: postgres
	title: PostgreSQL
	extras:
	- postgres
	- geospatial
	services:
	- postgres
	sys-deps:
	- libgeos-dev
	- os: windows-latest
	backend:
	name: risingwave
	title: Risingwave
	services:
	- risingwave
	extras:
	- risingwave
	- os: windows-latest
	backend:
	name: postgres
	title: PostgreSQL + Torch
	extras:
	- postgres
	- geospatial
	additional_deps:
	- torch
	services:
	- postgres
	sys-deps:
	- libgeos-dev
	- os: windows-latest
	backend:
	name: impala
	title: Impala
	serial: true
	extras:
	- impala
	services:
	- impala
	- kudu
	sys-deps:
	- cmake
	- ninja-build
	- os: windows-latest
	backend:
	name: mssql
	title: MS SQL Server
	extras:
	- mssql
	services:
	- mssql
	sys-deps:
	- freetds-dev
	- unixodbc-dev
	- tdsodbc
	- os: windows-latest
	backend:
	name: trino
	title: Trino
	services:
	- trino
	extras:
	- trino
	- os: windows-latest
	backend:
	name: druid
	title: Druid
	extras:
	- druid
	services:
	- druid
	- os: windows-latest
	backend:
	name: oracle
	title: Oracle
	serial: true
	extras:
	- oracle
	services:
	- oracle
	# - os: windows-latest
	# backend:
	# name: flink
	# title: Flink
	# serial: true
	# extras:
	# - flink
	# services:
	# - flink
	# - python-version: "3.11"
	# backend:
	# name: flink
	# title: Flink
	# serial: true
	# extras:
	# - flink
	# services:
	# - flink
	# - os: windows-latest
	# backend:
	# name: risingwave
	# title: Risingwave
	# services:
	# - risingwave
	# extras:
	# - risingwave
	- os: windows-latest
	backend:
	name: exasol
	title: Exasol
	serial: true
	extras:
	- exasol
	services:
	- exasol
	steps:
	- name: update and install system dependencies
	if: matrix.os == 'ubuntu-latest' && matrix.backend.sys-deps != null
	run: \|
	set -euo pipefail

	sudo apt-get update -qq -y
	sudo apt-get install -qq -y build-essential ${{ join(matrix.backend.sys-deps, ' ') }}

	- name: install sqlite
	if: matrix.os == 'windows-latest' && matrix.backend.name == 'sqlite'
	run: choco install sqlite

	- name: checkout
	uses: actions/checkout@v4

	- uses: extractions/setup-just@v1
	env:
	GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

	- name: download backend data
	run: just download-data

	- name: start services
	if: matrix.backend.services != null
	run: docker compose up --wait ${{ join(matrix.backend.services, ' ') }}

	- name: install poetry
	run: pipx install 'poetry==1.7.1'

	- name: install python
	uses: actions/setup-python@v5
	id: install_python
	with:
	python-version: ${{ matrix.python-version }}
	cache: poetry

	- name: install ibis
	run: poetry install --without dev --without docs --extras "${{ join(matrix.backend.extras, ' ') }}"

	- name: install deps for broken avro-python setup
	if: matrix.backend.name == 'flink'
	run: poetry run pip install wheel

	- name: install other deps
	if: matrix.backend.additional_deps != null
	run: poetry run pip install ${{ join(matrix.backend.additional_deps, ' ') }}

	- name: show installed deps
	run: poetry run pip list

	- name: "run parallel tests: ${{ matrix.backend.name }}"
	if: ${{ !matrix.backend.serial }}
	run: just ci-check -m ${{ matrix.backend.name }} --numprocesses auto --dist=loadgroup
	env:
	IBIS_TEST_IMPALA_HOST: localhost
	IBIS_TEST_IMPALA_PORT: 21050
	IBIS_EXAMPLES_DATA: ${{ runner.temp }}/examples-${{ matrix.backend.name }}-${{ matrix.os }}-${{ steps.install_python.outputs.python-version }}

	# FIXME(deepyaman): If some backend-specific test, in test_ddl.py,
	# executes before common tests, they will fail with:
	# org.apache.flink.table.api.ValidationException: Table `default_catalog`.`default_database`.`functional_alltypes` was not found.
	# Therefore, we run backend-specific tests second to avoid this.
	# - name: "run serial tests: ${{ matrix.backend.name }}"
	# if: matrix.backend.serial && matrix.backend.name == 'flink'
	# run: \|
	# just ci-check -m ${{ matrix.backend.name }} ibis/backends/tests
	# just ci-check -m ${{ matrix.backend.name }} ibis/backends/flink/tests
	# env:
	# IBIS_EXAMPLES_DATA: ${{ runner.temp }}/examples-${{ matrix.backend.name }}-${{ matrix.os }}-${{ steps.install_python.outputs.python-version }}
	# FLINK_REMOTE_CLUSTER_ADDR: localhost
	# FLINK_REMOTE_CLUSTER_PORT: "8081"
	#
	- name: "run serial tests: ${{ matrix.backend.name }}"
	if: matrix.backend.serial # && matrix.backend.name != 'flink'
	run: just ci-check -m ${{ matrix.backend.name }}
	env:
	IBIS_EXAMPLES_DATA: ${{ runner.temp }}/examples-${{ matrix.backend.name }}-${{ matrix.os }}-${{ steps.install_python.outputs.python-version }}

	- name: check that no untracked files were produced
	shell: bash
	run: git checkout poetry.lock pyproject.toml && ! git status --porcelain \| tee /dev/stderr \| grep .

	- name: upload code coverage
	if: success()
	continue-on-error: true
	uses: codecov/codecov-action@v4
	with:
	flags: backend,${{ matrix.backend.name }},${{ runner.os }},python-${{ steps.install_python.outputs.python-version }}

	- name: Show docker compose logs on fail
	if: matrix.backend.services != null && failure()
	run: docker compose logs

	test_backends_min_version:
	name: ${{ matrix.backend.title }} Min Version ${{ matrix.os }} python-${{ matrix.python-version }}
	runs-on: ${{ matrix.os }}
	strategy:
	fail-fast: false
	matrix:
	os:
	- ubuntu-latest
	- windows-latest
	python-version:
	- "3.9"
	- "3.11"
	backend:
	- name: dask
	title: Dask
	deps:
	- "dask[array,dataframe]@2022.9.1"
	- "[email protected]"
	extras:
	- dask
	- name: postgres
	title: PostgreSQL
	deps:
	- "[email protected]"
	- "[email protected]"
	- "Shapely@2"
	services:
	- postgres
	extras:
	- postgres
	- geospatial
	exclude:
	- os: windows-latest
	backend:
	name: postgres
	title: PostgreSQL
	deps:
	- "[email protected]"
	- "[email protected]"
	- "Shapely@2"
	services:
	- postgres
	extras:
	- postgres
	- geospatial
	- python-version: "3.11"
	backend:
	name: postgres
	title: PostgreSQL
	deps:
	- "[email protected]"
	- "[email protected]"
	- "Shapely@2"
	services:
	- postgres
	extras:
	- postgres
	- geospatial
	steps:
	- name: checkout
	uses: actions/checkout@v4

	- name: install libgeos for shapely
	if: matrix.backend.name == 'postgres'
	run: \|
	sudo apt-get update -y -qq
	sudo apt-get install -qq -y build-essential libgeos-dev

	- uses: extractions/setup-just@v1
	env:
	GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

	- name: download backend data
	run: just download-data

	- name: start services
	if: matrix.backend.services != null
	run: docker compose up --wait ${{ join(matrix.backend.services, ' ') }}

	- name: install python
	uses: actions/setup-python@v5
	id: install_python
	with:
	python-version: ${{ matrix.python-version }}

	- name: install poetry
	run: python -m pip install --upgrade pip 'poetry==1.7.1'

	- name: remove lonboard
	# it requires a version of pandas that min versions are not compatible with
	run: poetry remove lonboard

	- name: install minimum versions
	run: poetry add --lock --optional ${{ join(matrix.backend.deps, ' ') }}

	- name: checkout the lock file
	run: git checkout poetry.lock

	- name: lock with no updates
	# poetry add is aggressive and will update other dependencies like
	# numpy and pandas so we keep the pyproject.toml edits and then relock
	# without updating anything except the requested versions
	run: poetry lock --no-update

	- name: install ibis
	run: poetry install --without dev --without docs --extras "${{ join(matrix.backend.extras, ' ') }}"

	- name: run tests
	run: just ci-check -m ${{ matrix.backend.name }} --numprocesses auto --dist=loadgroup

	- name: check that no untracked files were produced
	shell: bash
	run: git checkout poetry.lock pyproject.toml && ! git status --porcelain \| tee /dev/stderr \| grep .

	- name: upload code coverage
	if: success()
	continue-on-error: true
	uses: codecov/codecov-action@v4
	with:
	flags: backend,${{ matrix.backend.name }},${{ runner.os }},python-${{ steps.install_python.outputs.python-version }}

	- name: Show docker compose logs on fail
	if: matrix.backend.services != null && failure()
	run: docker compose logs

	test_pyspark:
	name: PySpark ${{ matrix.pyspark-version }} ubuntu-latest python-${{ matrix.python-version }}
	runs-on: ubuntu-latest
	strategy:
	fail-fast: false
	matrix:
	include:
	- name: pyspark-3.3
	python-version: "3.9"
	pyspark-version: 3.3
	deps:
	- "[email protected]"
	- "'pandas@<2'"
	- "'numpy@<1.24'"
	- name: pyspark-3.5
	python-version: "3.11"
	pyspark-version: 3.5
	deps:
	- "[email protected]"
	- "'pandas@>2'"
	- "'numpy@>1.24'"
	steps:
	- name: checkout
	uses: actions/checkout@v4

	- uses: actions/setup-java@v4
	with:
	distribution: microsoft
	java-version: 17

	- uses: extractions/setup-just@v1
	env:
	GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

	- name: download backend data
	run: just download-data

	- name: install python
	uses: actions/setup-python@v5
	id: install_python
	with:
	python-version: ${{ matrix.python-version }}

	- name: install poetry
	run: python -m pip install --upgrade pip 'poetry==1.7.1'

	- name: remove lonboard
	# it requires a version of pandas that pyspark is not compatible with
	run: poetry remove lonboard

	- name: install exact versions of pyspark, pandas and numpy
	run: poetry add --lock ${{ join(matrix.deps, ' ') }}

	- name: checkout the lock file
	run: git checkout poetry.lock

	- name: lock with no updates
	# poetry add is aggressive and will update other dependencies like
	# numpy and pandas so we keep the pyproject.toml edits and then relock
	# without updating anything except the requested versions
	run: poetry lock --no-update

	- name: install ibis
	run: poetry install --without dev --without docs --extras pyspark

	- name: run tests
	run: just ci-check -m pyspark

	- name: check that no untracked files were produced
	shell: bash
	run: git checkout poetry.lock pyproject.toml && ! git status --porcelain \| tee /dev/stderr \| grep .

	- name: upload code coverage
	# only upload coverage for jobs that aren't mostly xfails
	if: success() && matrix.python-version != '3.11'
	continue-on-error: true
	uses: codecov/codecov-action@v4
	with:
	flags: backend,pyspark,${{ runner.os }},python-${{ steps.install_python.outputs.python-version }}

	backends:
	# this job exists so that we can use a single job from this workflow to gate merging
	runs-on: ubuntu-latest
	needs:
	# - test_backends_min_version
	- test_backends
	- test_pyspark
	steps:
	- run: exit 0

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix(backends): make string concat-with-null behavior consistent across backends #18990

Workflow file

fix(backends): make string concat-with-null behavior consistent across backends #18990

Jobs

Run details

Workflow file for this run