Hi Andrew, "Andrew Bachmann" <shatty@xxxxxxxxxxxxx> wrote: > > "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx> wrote: > > Update: at least the part of the glibc that already compiles as > > part of > > my tree works with these changes. > > But I am not sure if we want to replace gcc's stddef.h with our own > > at > > all. > This is an interesting patch actually. I needed to make this change > in order to configure and compile libiconv properly when using > our headers. AFAIK stddef.h is not part of the gcc distribution. It > is a posix header. It is included as part of gcc anyway because the > beos version of this header is broken, I think. If you look in the > /boot/develop/tools/gnupro/lib/gcc-lib/i586-pc-beos/.../include > directory you'll find these headers, along with a README that > says that these headers were generated from a script called > fixincludes. I couldn't find the fixincludes script but I did find > a script called fixproto which I may try to run against the OBOS > headers. It uses the "include_next" mechanism to patch headers - > sound familiar? Anyway, because I am sadistic, I may try to > compile 2.95.3 against the OBOS headers as well. That should be > enlightening. Yeah, do that. As I understand those fixing scripts, they are there to install GCC support on the host platform :-) And GCC seems to come with: stddef.h, stdarg.h, varargs.h, stdbool.h, and float.h. At least that's what I found in their current repository (at gcc/ginclude). But in the README-fixinc file, there is this paragraph: "Many of the files in this directory were automatically edited from the standard system header files by the fixincludes process. They are system-specific, and will not work on any other kind of system. They are also not part of GCC. The reason we have to do this is because GCC requires ANSI C headers and many vendors supply ANSI-incompatible headers." But looking in stddef.h I can hardly think that this is one of the headers that do not need a fix :-) Adios... Axel.