John Scipione wrote: > On Mon, Aug 13, 2012 at 6:16 AM, Ingo Weinhold <ingo_weinhold@xxxxxx> wrote: > > On 2012-08-13 at 02:34:04 [+0200], jscipione@xxxxxxxxx wrote: > >> hrev44524 adds 2 changesets to branch 'master' > >> old head: 7ad72ef6eae1ad8a2ec03043055ca0e649b11f80 > >> new head: efeef5fad7c8a92b0595540a6004113b8df8da12 > >> > >> ---------------------------------------------------------------------------- > >> > >> e4e6891: Cleanup darwin build system some more. > >> > >> * Convert fs_darwin.cpp and fs_freebsd.cpp to C. > > > > Please revert that. There's no reason for doing so and it prevents later C++ > > use. A good deal of the libroot_build implementation (particularly the > > attributes part) makes use of C++. > > The Haiku equivalent functions are in C. Is this because they are > included by the kernel? These functions are not going to be included > by the kernel (build system only) so I guess they should be C++? Kernel or not, there's no reason to force C compilation. It is easy to make C headers C++-safe. It is impossible to make C++ headers C safe (other than #ifdef'ing the functionality out that is). Moreover the C++ compiler is stricter (particularly wrt. types) which is a good thing, too. > I am nervous that I will get link errors because the functions are > declared with C linkage in the headers/build/host/<hostname>/header > files but the implementations in fs_<hostname> are C++. Is this > something that I should worry about? As long as the header declares the functions with C linkage (and the header is actually included) there's nothing to worry about. > Unfortunately I won't be able to convert back to C++ until tonight > since I need to test on OS X and FreeBSD. No hurry. CU, Ingo