We are working on packaging node for arm64 macs over at conda-forge
Attempting to cross-compile on mac, x86_64 -> arm64 with
./configure
--ninja \
--prefix=${PREFIX} \
--without-node-snapshot \
--shared \
--shared-libuv \
--shared-openssl \
--shared-zlib \
--with-intl=system-icu \
--cross-compiling --dest-os=mac --dest-cpu=arm64
fails with this error:
ninja: error: obj.host/tools/v8_gypfiles/run_torque.ninja:546: multiple rules generate gen/torque-output-root/torque-generated/bit-fields-tq.h [-w dupbuild=err]
I was able to narrow it down to the fact that v8/torque is built for both host and target, while SHARED_INTERMEDIATE_DIR is not unique per toolset.
Adding $|OBJ to SHARED_INTERMEDIATE_DIR via this patch so it is per-toolchain instead of per-product seems to work, but I don't know if that's actually the right fix. It seems plausible that these should only be generated once, and the bug really is is that they are being generated twice. I couldn't figure out how to solve that, though.
We are working on packaging node for arm64 macs over at conda-forge
Attempting to cross-compile on mac, x86_64 -> arm64 with
fails with this error:
I was able to narrow it down to the fact that v8/torque is built for both host and target, while SHARED_INTERMEDIATE_DIR is not unique per toolset.
Adding
$|OBJto SHARED_INTERMEDIATE_DIR via this patch so it is per-toolchain instead of per-product seems to work, but I don't know if that's actually the right fix. It seems plausible that these should only be generated once, and the bug really is is that they are being generated twice. I couldn't figure out how to solve that, though.