Re: Upstreaming the glibc Hurd port

[Zack Weinberg]

On Sun, Mar 18, 2018 at 9:51 PM, Samuel Thibault
<samuel.thibault@gnu.org> wrote:
> Hello,
>
> Thanks a lot for the feedback on what needs to be done.  It was a busy
> week-end :)
>
> We should be almost there, I believe I had addressed in
> sthibaul/hurd-builds all requirements except a few remaining bits:

Is it still expected that glibc trunk cannot be built in the
--host=i686-gnu configuration?  I am getting

In file included from ../include/link.h:52:0,
                 from ../include/dlfcn.h:4,
                 from ../sysdeps/generic/ldsodefs.h:32,
                 from ../sysdeps/i386/ldsodefs.h:39,
                 from ../sysdeps/gnu/ldsodefs.h:46,
                 from setup-thread.c:22:
../sysdeps/mach/libc-lock.h:227:11: fatal error: libc-lockP.h: No such
file or directory
 # include <libc-lockP.h>
           ^~~~~~~~~~~~~~
compilation terminated.

(I would like to suggest that getting "scripts/build-many-glibcs.py
$workdir compilers i686-gnu" to succeed, and
"scripts/build-many-glibcs.py $workdir glibcs i686-gnu" to fail only
in the testsuite phase, when $workdir/src/glibc contains a trunk
checkout, should be top priority -- once you get to that point, people
working on cross-cutting changes (like me) can ensure that they
haven't made the state of the Hurd port worse.)

zw