mirror of https://github.com/llvm/torch-mlir
Add tools/test_all.sh
We should look into having a `ninja check-npcomp` that runs everything with lit so that we get decent test multithreading. We can look to how LLVM does its gtest tests ("unittests") for inspiration.pull/1/head
parent
f2985e0901
commit
f394e12d86
|
@ -31,6 +31,8 @@ export LLVM_SRC_DIR=/path/to/llvm-project
|
|||
./tools/install_mlir.sh
|
||||
./tools/cmake_configure.sh
|
||||
|
||||
|
||||
# ./tools/test_all.sh runs all of these commands.
|
||||
cd build
|
||||
ninja
|
||||
ninja check-npcomp-opt
|
||||
|
|
|
@ -0,0 +1,31 @@
|
|||
#!/bin/bash
|
||||
|
||||
# Runs all the tests we are aware of in mlir-npcomp.
|
||||
# TODO: This should eventually all be folded into a `ninja check-npcomp`.
|
||||
# Unfortunately, that seems to be a level of build engineering that we
|
||||
# haven't had enough bandwidth for.
|
||||
#
|
||||
# The basic idea seems to be to teach `lit` how to run our Python tests.
|
||||
#
|
||||
# There is precendent for doing this, e.g.
|
||||
# - For GoogleTest tests (i.e. "unittests"): https://github.com/llvm/llvm-project/blob/c490c5e81ac90cbf079c7cee18cd56171f1e27af/llvm/test/Unit/lit.cfg.py#L25
|
||||
# - LLDB tests: https://github.com/llvm/llvm-project/blob/eb7d32e46fe184fdfcb52e0a25973e713047e305/lldb/test/API/lldbtest.py#L32
|
||||
# - libcxx tests: https://github.com/llvm/llvm-project/blob/eb7d32e46fe184fdfcb52e0a25973e713047e305/libcxx/utils/libcxx/test/newformat.py#L192
|
||||
#
|
||||
# The `getTestsInDirectory` keyword seems to be what to search for.
|
||||
|
||||
set -euo pipefail
|
||||
td="$(realpath $(dirname $0)/..)"
|
||||
|
||||
cd $td/build
|
||||
|
||||
ninja
|
||||
ninja check-npcomp-opt
|
||||
./python/run_tests.py
|
||||
|
||||
echo
|
||||
echo "========"
|
||||
echo "ALL PASS"
|
||||
echo "========"
|
||||
|
||||
|
Loading…
Reference in New Issue