bobbin-wasm leverages the wabt test suite by implementing binaries namded
wasm-objdump and wasm-interp that generate output byte-for-byte identical to their wabt counterparts.
Download and install the current version of wabt, and make a symlink from ./test/ to wabt/test.
Create a local directory ./bin/.
# mkdir ./bin/
Add symlinks for all of the wabt binaries to the ./bin directory:
ln -s path_to_wabt/bin/ bin/
Build and install the wasm-objdump and wasm-interp binaries into your test-bin directory.
$ cargo install --path . --root .
To run a subset of the test suite, use make test. You can also run make test-dump or make test-interp to
run just the wasm-objdump or wasm-interp tests.
You should see something like this as the result:
$ make test
cargo -q install --path . --root . --force
cat test-dump.txt | grep -v \# | xargs test/run-tests.py --bindir /Users/jcsoo/.cargo/bin
[+71|-0|%100] (0.96s) test/dump/unreachable.txt
cat test-interp.txt | grep -v \# | xargs test/run-tests.py --bindir /Users/jcsoo/.cargo/bin
[+19|-0|%100] (0.18s) test/interp/unreachable.txt
$
bobbin-wasm doesn't currently support all the functionality of the real versions of wasm-objdump and wasm-interp, so
testing them against the full test suite would produce an unreasonable number of failures.
Instead, the files test-dump.txt and test-interp.txt list the names of the tests that should be run for each tool. Tests that are known to fail are commented out with a hash mark in the first column.
Additionally, local_test contains copies of tests that have been useful during development. Some of these are tests
from the wabt test suite that have had specific unsupported functionality removed. There is not currently a way to
run these tests in an automated fashion.