Regression Testing: Use Mainline & CMake

.azure-pipelines.yml

@@ -13,6 +13,7 @@ jobs:
     BLASPP_HOME: '/usr/local'
     CEI_SUDO: 'sudo'
     CEI_TMP: '/tmp/cei'
+    CMAKE_GENERATOR: 'Ninja'
     FFTW_HOME: '/usr'
     LAPACKPP_HOME: '/usr/local'
     OMP_NUM_THREADS: 1
@@ -75,7 +76,8 @@ jobs:
       set -eu -o pipefail
       cat /proc/cpuinfo | grep "model name" | sort -u
       df -h
-      sudo apt install -y ccache curl gcc gfortran git g++ openmpi-bin libopenmpi-dev \
+      sudo apt install -y ccache curl gcc gfortran git g++ ninja-build \
+        openmpi-bin libopenmpi-dev \
         libfftw3-dev libfftw3-mpi-dev libhdf5-openmpi-dev pkg-config make \
         python3 python3-pip python3-venv python3-setuptools libblas-dev liblapack-dev
       ccache --set-config=max_size=10.0G

CMakeLists.txt

@@ -391,6 +391,25 @@ endif()
 #)
 
 
+# CI Helper ###################################################################
+#
+
+# calling `python -m pip wheel .` and install re-using the build directory
+# note: this is mainly for our CI Python regression scripts, users and package
+#       managers should just use `python -m pip wheel .` and `... install *whl`
+#       directly
+set(PYINSTALLOPTIONS "" CACHE STRING "Additional parameters to pass to `pip install`")
+add_custom_target(install_pip
+    ${CMAKE_COMMAND} -E env PYWARPX_LIB_DIR=${CMAKE_LIBRARY_OUTPUT_DIRECTORY} python3 -m pip wheel -v --use-feature=in-tree-build ${WarpX_SOURCE_DIR}
+    COMMAND
+        python3 -m pip install --force-reinstall -v ${PYINSTALLOPTIONS} ${CMAKE_BINARY_DIR}/*whl
+    WORKING_DIRECTORY
+        ${CMAKE_BINARY_DIR}
+    DEPENDS
+        shared
+)
+
+
 # Tests #######################################################################
 #
 

Docs/source/developers/testing.rst

@@ -89,11 +89,12 @@ There are three steps to follow to add a new automated test (illustrated here fo
    runtime_params = warpx.do_dynamic_scheduling=0 algo.maxwell_solver=yee
    dim = 2
    addToCompileString =
+   cmakeSetupOpts = -DWarpX_DIMS=2
    restartTest = 0
    useMPI = 1
    numprocs = 2
    useOMP = 1
-   numthreads = 2
+   numthreads = 1
    compileTest = 0
    doVis = 0
    analysisRoutine = Examples/Tests/PML/analysis_pml_yee.py
@@ -104,8 +105,14 @@ If you re-use an existing input file, you can add arguments to ``runtime_params`
 
    If you added ``analysisRoutine = Examples/analysis_default_regression.py``, then run the new test case locally and add the :ref:`checksum <developers-checksum>` file for the expected output.
 
+.. note::
+
+   We run those tests on our continuous integration services, which at the moment only have 2 virtual CPU cores.
+   Thus, make sure that the product of ``numprocs`` and ``numthreads`` for a test is ``<=2``.
+
+
 Useful tool for plotfile comparison: ``fcompare``
---------------------------------------------------
+-------------------------------------------------
 
 AMReX provides ``fcompare``, an executable that takes two ``plotfiles`` as input and returns the absolute and relative difference for each field between these two plotfiles. For some changes in the code, it is very convenient to run the same input file with an old and your current version, and ``fcompare`` the plotfiles at the same iteration. To use it:
 

Examples/Modules/laser_injection_from_file/analysis.py

@@ -218,7 +218,7 @@ def launch_analysis(executable):
 
 
 def main() :
-    executables = glob.glob("main2d*")
+    executables = glob.glob("*.ex")
     if len(executables) == 1 :
         launch_analysis(executables[0])
     else :

Regression/WarpX-GPU-tests.ini

@@ -19,6 +19,10 @@ purge_output = 1
 MAKE = make
 numMakeJobs = 8
 
+# We build by default a few tools for output comparison.
+# The build time for those can be skipped if they are not needed.
+ftools =
+
 # Control the build of the particle_compare tool.
 # Needed for test particle_tolerance option.
 use_ctools = 0