Disable cache on the CI (deepset-ai#2083)

* Disable cache on the CI

* Reintroduce paths

* Add most files to the cache key

* remove date and path from cache key

* Try double install with cache

* Try to cache more stuff, on a per-commit basis

* Fix windows CI too

* Add comment on how to speed up the CI with better caching

.github/workflows/linux_ci.yml

@@ -28,18 +28,23 @@ jobs:
  - uses: actions/setup-python@v2
  with:
  python-version: 3.7
- - run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_ENV
+
  - name: Cache
  id: cache-python-env
  uses: actions/cache@v2
  with:
  path: ${{ env.pythonLocation }}
- key: linux-${{ env.pythonLocation }}-${{ env.date }}-${{ hashFiles('setup.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+ # The cache currently misses at every change in the Python files to ensure
+ # that the code changes to Haystack are not being cached along with the dependencies.
+ # TODO for the tests refactoring: We could speed up the process by caching only 
+ # the dependencies and installing Haystack again separately for each test runner.
+ key: linux-${{ hashFiles('**/*.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+
  - name: Install dependencies
  if: steps.cache-python-env.outputs.cache-hit != 'true'
  run: |
- python -m pip install --upgrade pip
- pip install --upgrade --upgrade-strategy eager -e .[test]
+ pip install --upgrade pip
+ pip install --upgrade --upgrade-strategy eager .[test]
  pip install torch-scatter -f https://data.pyg.org/whl/torch-1.10.0+cpu.html
 
  prepare-build:
@@ -52,6 +57,7 @@ jobs:
  echo "::set-output name=matrix::$(find $(find . -type d -name test -not -path "./*env*/*") -type f -name test_*.py | jq -SR . | jq -cs .)"
  outputs:
  matrix: ${{ steps.set-matrix.outputs.matrix }}
+
  build:
  needs: prepare-build
  runs-on: ubuntu-20.04
@@ -61,16 +67,18 @@ jobs:
  fail-fast: false
  steps:
  - uses: actions/checkout@v2
+
  - name: Set up Python 3.7
  uses: actions/setup-python@v2
  with:
  python-version: 3.7
- - run: echo "date=$(date +'%Y-%m-%d')" >> $GITHUB_ENV
- - name: Cache
+
+ - name: Cache Python
  uses: actions/cache@v2
  with:
  path: ${{ env.pythonLocation }}
- key: linux-${{ env.pythonLocation }}-${{ env.date }}-${{ hashFiles('setup.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+ key: linux-${{ hashFiles('**/*.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+
  - name: Run Elasticsearch
  run: docker run -d -p 9200:9200 -e "discovery.type=single-node" -e "ES_JAVA_OPTS=-Xms128m -Xmx128m" elasticsearch:7.9.2
 

.github/workflows/windows_ci.yml

@@ -29,21 +29,27 @@ jobs:
  - uses: actions/setup-python@v2
  with:
  python-version: 3.7
- - run: echo "date=$(date +'%Y-%m-%d')" >> $env:GITHUB_ENV
+
  - name: Cache
  id: cache-python-env
  uses: actions/cache@v2
  with:
  path: ${{ env.pythonLocation }}
- key: windows-${{ env.pythonLocation }}-${{ env.date }}-${{ hashFiles('setup.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+ # The cache currently misses at every change in the Python files to ensure
+ # that the code changes to Haystack are not being cached along with the dependencies.
+ # TODO for the tests refactoring: We could speed up the process by caching only 
+ # the dependencies and installing Haystack again separately for each test runner.
+ key: windows-${{ hashFiles('**/*.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+
  - name: Install Pytorch on windows
  run: |
  pip install torch==1.8.1+cpu -f https://download.pytorch.org/whl/lts/1.8/torch_lts.html
+ 
  - name: Install dependencies
  if: steps.cache-python-env.outputs.cache-hit != 'true'
  run: |
  python -m pip install --upgrade pip
- pip install --upgrade --upgrade-strategy eager -e .[test]
+ pip install --upgrade --upgrade-strategy eager .[test]
  pip install torch-scatter -f https://data.pyg.org/whl/torch-1.10.0+cpu.html
 
  prepare-build:
@@ -71,12 +77,12 @@ jobs:
  uses: actions/setup-python@v2
  with:
  python-version: 3.7
- - run: echo "date=$(date +'%Y-%m-%d')" >> $env:GITHUB_ENV
+
  - name: Cache
  uses: actions/cache@v2
  with:
  path: ${{ env.pythonLocation }}
- key: windows-${{ env.pythonLocation }}-${{ env.date }}-${{ hashFiles('setup.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
+ key: windows-${{ hashFiles('**/*.py') }}-${{ hashFiles('setup.cfg') }}-${{ hashFiles('pyproject.toml') }}
 
  # Windows runner can't run Linux containers. Refer https://github.com/actions/virtual-environments/issues/1143
  - name: Set up Windows test env

haystack/__init__.py

@@ -9,6 +9,7 @@
 
 __version__ = metadata.version('farm-haystack')
 
+
 # This configuration must be done before any import to apply to all submodules
 import logging
 logging.basicConfig(format="%(levelname)s - %(name)s - %(message)s", datefmt="%m/%d/%Y %H:%M:%S", level=logging.WARNING)

setup.cfg

@@ -48,6 +48,9 @@ classifiers =
 use_scm_version = True
 python_requires = >=3.7
 packages = find:
+setup_requires =
+ setuptools
+ wheel
 install_requires =
  importlib-metadata; python_version < '3.8'
  torch>1.9,<1.11