All libffi tests pass now. The only failing ctypes test is test_callback_too_many_args which doesn't segfault anymore, it only soft fails. Planning to submit a PR to cpython that fixes test_callback_too_many_args.
See also:
bugs.python.org/issue47208
https://github.com/emscripten-core/emscripten/pull/16658
For reasons that are a bit beyond me, `--host` and `PLATFORM_TRIPLET`
seem to be independent, in particular we've had an empty
`PLATFORM_TRIPLET`. This is unfortunate because `PLATFORM_TRIPLET`
is used to generate the SOABI config variable which in turn is used
to decide whether a .so file is a good match for loading. We'd like
for linux Pythons not to try to import emscripten .so files (it
raises `ImportError: some_file.so: invalid ELF header`). Similarly,
we'd like to avoid attempting to load linux .so files in wasm. These
platform tags are our friends.
Anyways, this PR sets `PLATFORM_TRIPLET` and ensures that .so files
built by pywasmcross are tagged with our SOABI tag.
I moved the .so file renaming from pywasmcross to buildpkg just
before running the post script. That is a better place to put it in
case the package wants to look at the .so file after linking it. It
might be surprised that we moved it.
I also improved the error message if we try to `loadWebAssemblyModule`
something that is actually say a Linux .so file and updated get_dynlibs
to filter out .so files that have an incompatible abi tag.
The goal of this commit is to add the capability of skipping specific tests by
glob pattern. Many core test sets have just one test that uses subprocess or
threads. Others have a few tests that are failing due to upstream problems in
emscripten. I went through and switched to skipping specific tests in as many
cases as I could. In modules where most or all tests use threading or
subprocess, I gave up and put a comment saying something like 35/50 tests fork
and I didn't want to bother to salvage the remaining ones.
In order to accommodate using patterns to skip specific tests (and also extra
comments) I switched the format of python_tests to a yaml file. This gives us
significantly more flexibility in the structure of the file and allows us to use
ruamel.yaml to update it. We gain a lot flexibility while reducing the length of
make_test_lists.py modestly from 108 to 78 loc.
In the future, hopefully we should be able to make fixes by deleting one skip
pattern and looking into why that particular test fails.
* Patch ceval to add interrupt handler callback
* Don't use SharedArrayBuffer in test, Firefox doesn't expose it
* Fix test
* Retest
Co-authored-by: Jan Max Meyer <jmm@phorward.de>
This is needed for the upstream backend, and I believe is a noop under
the fastcomp backend
Co-authored-by: Michael Droettboom <mdboom@gmail.com>
Co-authored-by: Joe Marshall <joe.marshall@nottingham.ac.uk>
In python's Makefile generated by makesetup, SQLITEBUILD and BZIP2BUILD
are not defined, so we are left with empty -I and -L commands. These
manage to build by virtue of us inserting -I/path/to/sqlite and
-I/path/to/bzip2 to every single build command.
This commit does the same for zlib for consistency.
When sqlite was originally added, the issue
described was bypassed by patching the python files. An alternative
workaround is used here that doesn't require any patching.
Ryan May