I'm trying to build a fftw dynamic library for macOS/x86 on a macOS/arm system. I am able to build a dynamic library for arm and a static library for x86, but unable to build a dynamic library for x86.
This code builds a static library for x86: export CFLAGS="-arch x86_64" ./configure --host=x86_64-apple-macos10.10 --target=x86_64-apple-macos10.10 This code should also build a dynamic library, but it does not: export CFLAGS="-arch x86_64" ./configure --host=x86_64-apple-macos10.10 --target=x86_64-apple-macos10.10 --enable-shared The configure log says some unexpected things: checking whether the gcc linker (/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/ld) supports shared libraries... no checking if libtool supports shared libraries... no When building for arm (i.e., not cross compiling), these say yes. I’m a complete gnub when it comes to autoconf and similar tools, but it looks like this could be a bug. I also noticed something odd in the configure log: configure:4614: checking whether we are cross compiling configure:4622: gcc -o conftest -arch x86_64 conftest.c >&5 configure:4626: $? = 0 configure:4633: ./conftest configure:4637: $? = 0 configure:4652: result: no It fails to recognize cross compilation because the cross compiled executable actually runs on macOS. I don’t know if this is related. Just out of curiosity, why do I need to set CFLAGS? Doesn’t the host and target provide enough information?
