On 2010-11-04 at 19:47:57 [+0100], François Revol <revol@xxxxxxx> wrote: > Le 4 nov. 2010 à 19:08, Ingo Weinhold a écrit : > > On 2010-11-04 at 17:30:01 [+0100], François Revol <revol@xxxxxxx> wrote: > >> Le 4 nov. 2010 à 17:26, ingo_weinhold@xxxxxx a écrit : > >> > >>> Author: bonefish > >>> Date: 2010-11-04 17:26:43 +0100 (Thu, 04 Nov 2010) > >>> New Revision: 39300 > >>> Changeset: http://dev.haiku-os.org/changeset/39300 > >>> > >>> Modified: > >>> haiku/trunk/src/system/kernel/util/kernel_cpp.cpp > >>> Log: > >>> Added dummy __cxa_atexit() and __cxa_finalize() to the kernel. We might > >>> actually implement those, if we decide to support global > >>> constructors/destructors in kernel add-ons. > >> > >> Shouldn't it at least dprintf() something to avoid searching for hours > >> why > >> the code is not called ? > > > > Since we don't even call the function that would call __cxa_finalize(), > > there's little point in adding a dprintf(). > > This one no, but maybe the former, from ported code, or if someone tries to > put a global and ask why it doesn't work. Unless I'm much mistaken __cxa_atexit() is 1. only used for C++ and 2. only for function static objects. Since we don't even support static global constructors/destructors you'll probably run into other problems first. CU, Ingo