cvs commit: src/sys/sys cdefs.h src/include nl_types.h stdio.h
ru at FreeBSD.org
Wed Feb 1 03:30:31 PST 2006
On Wed, Feb 01, 2006 at 09:02:58AM +0100, Stefan Farfeleder wrote:
> On Tue, Jan 31, 2006 at 01:57:34PM -0700, M. Warner Losh wrote:
> > Confirmed. RELENG_5 -> RELENG_6 upgrades are now broken.
> > ===> usr.bin/gencat (obj,depend,all,install)
> > In file included from /u2/imp/FreeBSD/6x/src/usr.bin/gencat/gencat.c:88:
> > /u2/imp/obj/u2/imp/FreeBSD/6x/src/tmp/legacy/usr/include/nl_types.h:100: error: syntax error before "__format_arg"
> > *** Error code 1
> > The fix I have applied locally is:
> > @@ -97,6 +97,9 @@
> > __BEGIN_DECLS
> > nl_catd catopen(const char *, int);
> > +#ifndef __format_arg
> > +#define __format_arg(a) __attribute__((__format_arg__(a)))
> > +#endif
> > char *catgets(nl_catd, int, int, const char *) __format_arg(4);
> > int catclose(nl_catd);
> > __END_DECLS
> > although I'm sure that there's many ways to skin this cat...
> This patch kind of defeats the purpose of my commit. Why is gencat
> picking up the new <nl_tpes.h> but the old <sys/cdefs.h>? Shouldn't there
> be a "-I /u2/[...]/src/tmp/legacy/usr/include/sys" somewhere?
Because attempting to use a new cdefs.h will likely break it in
other way. This tool is compiled from fresh sources but using
the currently available toolchain because it's compiled to be run
on the host/OS/version doing buildworld, and it meant to be run
> Anyway this probably won't be fixed very soon, so should I commit something
> like your patch or back out the whole thing?
Commit what Warner sent you, but only to RELENG_6.
ru at FreeBSD.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/cvs-src/attachments/20060201/207f7fb2/attachment.bin
More information about the cvs-src