I have a large C/C++ library that I need to use as part of an Android NDK project. This library needs to be able to intelligently process UTF8 strings (for example, convers
We are using ICU in the NDK. Follow the steps in the ICU cross building instructions and you'll be OK. Basically you'll have a ICU native directory (for example Windows or Linux), a ICU Cygwin (if using such) and yet another for ICU Android (ARM). Sounds crazy, but it works!
Here are the steps for building under Cygwin. I am using the 'CrystaX' NDK r4, but it should build with the out of the box NDK as well. ICU version 4.4, but has worked with previous versions as well.
Patches:
Build your host version of ICU (e.g. Windows) as per normal. (We'll call this $HOST_ICU)
Build Cygwin ICU:
Build NDK version of ICU:
I use something like this for passed in (to step #4) CPPFLAGS/CXXFLAGS/CFLAGS:
-I$NDK_ROOT/build/platforms/android-8/arch-arm/usr/include/ -O3 -fno-short-wchar -DU_USING_ICU_NAMESPACE=0 -DU_GNUC_UTF16_STRING=0 -fno-short-enums -nostdlib
And for LDFLAGS:
-lc -Wl,-rpath-link=$NDK_ROOT/build/platforms/android-8/arch-arm/usr/lib/ -L$NDK_ROOT/build/platforms/android-8/arch-arm/usr/lib/
And additional configure params:
--enable-extras=no --enable-strict=no --enable-static --enable-shared=no --enable-tests=no --enable-samples=no --enable-dyload=no --enable-tools=no --host=arm-eabi --with-data-packaging=archive
I haven't done this manually for a while, it's currently all in a custom Python based build script. If you run into any other troubles, I can probably tell you what the issue is.
Good luck!