The runners on emulated hardware take significantly longer than the ones on native hardware - especially when building from a cold ccache.
The main point of those runners is to check if there are platform-specific errors. It is probably not actually necessary to build most of the objects twice (for the static and the shared version of each library) to check that.
To cut down on the build time, build only shared libraries on the (emulated) Alpine runners.
Additionally, this can serve as a check if there are issues with that configuration (i.e., building shared libraries only).
The runners on emulated hardware take significantly longer than the ones on native hardware - especially when building from a cold ccache. The main point of those runners is to check if there are platform-specific errors. It is probably not actually necessary to build most of the objects twice (for the static and the shared version of each library) to check that.
To cut down on the build time, build only shared libraries on the (emulated) Alpine runners.
Additionally, this can serve as a check if there are issues with that configuration (i.e., building shared libraries only).