[cmake-developers] Question about cross-compilation concept in CMake

[Jiri Malak]

Hi,

I am working on full cross-compilation stuff for Open Watcom toolchain in CMake.
I study CMake internals, but I am not sure what is prefered method for cross-compilation by CMake.
I see two concept one for Microsoft compilers where each target platform has own generator and
second one is for UNIX like toolchains which switch prefix/suffix to names for one architecture
per taget platform.

Open Watcom support 16-bit and 32-bit Intel processor architecture for each target platform (DOS,
Windows, OS/2) and run on DOS, OS/2(16/32), Windows(16/32/64bit), Linux(32/64bit) hosts.

For Open Watcom looks better first approach, define one Generator for each processor architecture,
now 16 and 32-bit Intel processors, next for 64-bit intel processor (as soon as OW will support
it) and define one target platform for each OS type.

Please, can you explain me what is preferred approach for CMake future.

Thanks

Jiri