Hi, Ingo Weinhold a écrit :
On 2007-10-08 at 00:19:48 [+0200], Olivier Coursiere <olivier.coursiere@xxxxxxxxxxx> wrote:The problem is solved. About one week ago i have configured the bone target and run "jam". Then i have configured my working copy to haiku last weekend. generated/objects/bone/x86/common/tools/gensyscalls is searched first. And because of "jam" on the Bone target, there was an older version of the syscalls.* files in this directory. I have removed those files, and now it compile again !I have a problem building latest revisions of Haiku. Here is the first error with revision 22478 : C++ generated/objects/bone/x86/release/tools/gensyscalls/gensyscalls_infos.o /BeOS2/tmp/haiku/generated/objects/haiku/x86/common/tools/gensyscalls/gensys calls_infos.cpp: In function `class SyscallVector * create_syscall_vector()': /BeOS2/tmp/haiku/generated/objects/haiku/x86/common/tools/gensyscalls/gensys calls_infos.cpp:39: `object_wait_info' undeclared (first use this function)The object_wait_info structure is defined in headers/os/kernel/OS.h, which is included from headers/private/kernel/syscalls.h. Can you please have a look at the preprocessed version of this header (generated/objects/haiku/x86/common/tools/gensyscalls/syscalls.h.pp), which is included by gensyscalls_infos.cpp, and check what goes wrong.CU, IngoPS: If you're using the StatCacheServer it sometimes happens, that it misses a node monitoring message (e.g. during an "svn up") and jam doesn't rebuild a dependent target, because it thinks the file is still up to date. Restarting the StatCacheServer fixes the problem in such a situation.
Next time i use the bone target, i should compile only the needed part instead of a global "jam".
Thanks for your hints. Olivier