Test failing in Cirrus CI on macos arm64
by Andy Alt from LinuxQuestions.org on (#6J0M4)
I'm working on NetPanzer and we have tests using GitHub Actions and Cirrus. They pass on all platforms but macos arm64 in the Cirrus CI.
Where it fails is when the two NetPanzer tests are run. It fails to match the data directory with the source root (where the data directories reside).
When NETPANZER_DATADIR is set, it's used for the datadir. The test just basically matches the environmental variable with MESON_SOURCE_ROOT, which is defined in tests/meson.build.
Code: Package::assignDataDir();
fprintf(stderr, "datadir: %s\n", Package::getDataDir().c_str());
assert(Package::getDataDir() == MESON_SOURCE_ROOT);The result on macos arm64 (Cirrus CI):
Quote:
So meson for some reason is prepending '/private' to the path.
datadir should match the env var CIRRUS_WORKING_DIR, which I've used in .cirrus.yml, which it does.
This almost sounds like a meson glitch, but figured I'd check here first since I have absolutely no experience with arm arches and little with OS X.
I can rewrite the test a little to account for this, or try on some other CI service, but I wanted to try to get a little more info about possible causes first.
Where it fails is when the two NetPanzer tests are run. It fails to match the data directory with the source root (where the data directories reside).
When NETPANZER_DATADIR is set, it's used for the datadir. The test just basically matches the environmental variable with MESON_SOURCE_ROOT, which is defined in tests/meson.build.
Code: Package::assignDataDir();
fprintf(stderr, "datadir: %s\n", Package::getDataDir().c_str());
assert(Package::getDataDir() == MESON_SOURCE_ROOT);The result on macos arm64 (Cirrus CI):
Quote:
| datadir: /var/folders/n8/pp_chp6x3jlfy1n94r6hdjv40000gn/T/cirrus-ci-build meson_source_root: /private/var/folders/n8/pp_chp6x3jlfy1n94r6hdjv40000gn/T/cirrus-ci-build |
datadir should match the env var CIRRUS_WORKING_DIR, which I've used in .cirrus.yml, which it does.
This almost sounds like a meson glitch, but figured I'd check here first since I have absolutely no experience with arm arches and little with OS X.
I can rewrite the test a little to account for this, or try on some other CI service, but I wanted to try to get a little more info about possible causes first.