Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Robert Pluim]

>>>>> On Sun, 18 Jun 2023 20:56:09 -0400 (EDT), Po Lu via Mailing list for Emacs changes <emacs-diffs@gnu.org> said:

    Po Lu> branch: feature/android
    Po Lu> commit 83eeebe9ef21e7278f5da3c233171f60bdf07cee
    Po Lu> Merge: 273c65f9e90 edb0862f5e6
    Po Lu> Author: Po Lu <luangruo@yahoo.com>
    Po Lu> Commit: Po Lu <luangruo@yahoo.com>

    Po Lu>     Merge remote-tracking branch 'origin/master' into feature/android

I doubt itʼs because of this specific change, but this branch doesnʼt
build on GNU/Linux, presumably because gnulib is now substituting
vfprintf which it wasnʼt before:

make -C lib-src all
make[2]: Entering directory '/home/rpluim/repos/emacs-android/lib-src'
  CCLD     seccomp-filter
/usr/bin/ld: /tmp/cckKZx81.o: in function `fail':
/home/rpluim/repos/emacs-android/lib-src/seccomp-filter.c:72: undefined reference to `rpl_vfprintf'
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile:483: seccomp-filter] Error 1
make[2]: Leaving directory '/home/rpluim/repos/emacs-android/lib-src'
make[1]: *** [Makefile:537: lib-src] Error 2
make[1]: Leaving directory '/home/rpluim/repos/emacs-android'

Robert
-- 

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Po Lu]

Robert Pluim <rpluim@gmail.com> writes:

>>>>>> On Sun, 18 Jun 2023 20:56:09 -0400 (EDT), Po Lu via Mailing list for Emacs changes <emacs-diffs@gnu.org> said:
>
>     Po Lu> branch: feature/android
>     Po Lu> commit 83eeebe9ef21e7278f5da3c233171f60bdf07cee
>     Po Lu> Merge: 273c65f9e90 edb0862f5e6
>     Po Lu> Author: Po Lu <luangruo@yahoo.com>
>     Po Lu> Commit: Po Lu <luangruo@yahoo.com>
>
>     Po Lu>     Merge remote-tracking branch 'origin/master' into feature/android
>
> I doubt itʼs because of this specific change, but this branch doesnʼt
> build on GNU/Linux, presumably because gnulib is now substituting
> vfprintf which it wasnʼt before:
>
> make -C lib-src all
> make[2]: Entering directory '/home/rpluim/repos/emacs-android/lib-src'
>   CCLD     seccomp-filter
> /usr/bin/ld: /tmp/cckKZx81.o: in function `fail':
> /home/rpluim/repos/emacs-android/lib-src/seccomp-filter.c:72: undefined reference to `rpl_vfprintf'
> collect2: error: ld returned 1 exit status
> make[2]: *** [Makefile:483: seccomp-filter] Error 1
> make[2]: Leaving directory '/home/rpluim/repos/emacs-android/lib-src'
> make[1]: *** [Makefile:537: lib-src] Error 2
> make[1]: Leaving directory '/home/rpluim/repos/emacs-android'
>
> Robert

Does this fix the problem?

diff --git a/lib-src/Makefile.in b/lib-src/Makefile.in
index 8a1922703de..8a5ce019ca2 100644
--- a/lib-src/Makefile.in
+++ b/lib-src/Makefile.in
@@ -481,7 +481,7 @@ emacsclient.res:
 ifeq ($(SECCOMP_FILTER),1)
 seccomp-filter$(EXEEXT): $(srcdir)/seccomp-filter.c $(config_h)
 	$(AM_V_CCLD)$(CC) $(ALL_CFLAGS) $(LIBSECCOMP_CFLAGS) $< \
-	  $(LIBSECCOMP_LIBS) -o $@
+	  $(LIBSECCOMP_LIBS) $(LOADLIBES) -o $@
 
 seccomp-filter.bpf seccomp-filter.pfc seccomp-filter-exec.bpf seccomp-filter-exec.pfc: seccomp-filter$(EXEEXT)
 	$(AM_V_GEN)./seccomp-filter$(EXEEXT) \

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Robert Pluim]

>>>>> On Mon, 19 Jun 2023 16:18:31 +0800, Po Lu <luangruo@yahoo.com> said:

    Po Lu> Does this fix the problem?

Yes, although the resulting emacs now interposes rpl_vfprintf and rpl_printf
instead of the glibc functions. Why is that suddenly necessary?

Robert
-- 

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Po Lu]

Robert Pluim <rpluim@gmail.com> writes:

>>>>>> On Mon, 19 Jun 2023 16:18:31 +0800, Po Lu <luangruo@yahoo.com> said:
>
>     Po Lu> Does this fix the problem?
>
> Yes, although the resulting emacs now interposes rpl_vfprintf and rpl_printf
> instead of the glibc functions. Why is that suddenly necessary?

Because Gnulib has detected that [vasn]printf aren't POSIX or C23
compliant.  It's not possible to avoid using Gnulib replacement
functions in lib-src, as long as the headers there include config.h.

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Robert Pluim]

>>>>> On Mon, 19 Jun 2023 16:41:28 +0800, Po Lu <luangruo@yahoo.com> said:

    Po Lu> Robert Pluim <rpluim@gmail.com> writes:
    >>>>>>> On Mon, 19 Jun 2023 16:18:31 +0800, Po Lu <luangruo@yahoo.com> said:
    >> 
    >> Po Lu> Does this fix the problem?
    >> 
    >> Yes, although the resulting emacs now interposes rpl_vfprintf and rpl_printf
    >> instead of the glibc functions. Why is that suddenly necessary?

    Po Lu> Because Gnulib has detected that [vasn]printf aren't POSIX or C23
    Po Lu> compliant.  It's not possible to avoid using Gnulib replacement
    Po Lu> functions in lib-src, as long as the headers there include config.h.

But master doesnʼt do that. Or is this branch using a newer version of
gnulib?

Robert
-- 

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Mon, 19 Jun 2023 10:51:06 +0200
> 
> >>>>> On Mon, 19 Jun 2023 16:41:28 +0800, Po Lu <luangruo@yahoo.com> said:
> 
>     Po Lu> Robert Pluim <rpluim@gmail.com> writes:
>     >>>>>>> On Mon, 19 Jun 2023 16:18:31 +0800, Po Lu <luangruo@yahoo.com> said:
>     >> 
>     >> Po Lu> Does this fix the problem?
>     >> 
>     >> Yes, although the resulting emacs now interposes rpl_vfprintf and rpl_printf
>     >> instead of the glibc functions. Why is that suddenly necessary?
> 
>     Po Lu> Because Gnulib has detected that [vasn]printf aren't POSIX or C23
>     Po Lu> compliant.  It's not possible to avoid using Gnulib replacement
>     Po Lu> functions in lib-src, as long as the headers there include config.h.
> 
> But master doesnʼt do that. Or is this branch using a newer version of
> gnulib?

Can you show the failed test program and the error messages?  Maybe we
should take this up with the Gnulib developers.  AFAIU, modern
GNU/Linux systems which use new enough glibc aren't supposed to
require replacements of library functions.

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

> Can you show the failed test program and the error messages?  Maybe we
> should take this up with the Gnulib developers.  AFAIU, modern
> GNU/Linux systems which use new enough glibc aren't supposed to
> require replacements of library functions.

No need; the Gnulib manual says:

  This function does not support the ‘b’ directive, required by ISO C23,
  on some platforms: glibc 2.34, musl libc, macOS 12.5, FreeBSD 13.1,
  NetBSD 9.0, OpenBSD 7.2, AIX 7.2, Solaris 11.4, Cygwin 2.9.0.

Glibc 2.37 does support this directive; Robert's Glibc is likely too
old.

Re: feature/android 83eeebe9ef2: Merge remote-tracking branch 'origin/master' into feature/android

[Robert Pluim]

>>>>> On Tue, 20 Jun 2023 08:54:06 +0800, Po Lu <luangruo@yahoo.com> said:

    Po Lu> Eli Zaretskii <eliz@gnu.org> writes:
    >> Can you show the failed test program and the error messages?  Maybe we
    >> should take this up with the Gnulib developers.  AFAIU, modern
    >> GNU/Linux systems which use new enough glibc aren't supposed to
    >> require replacements of library functions.

    Po Lu> No need; the Gnulib manual says:

    Po Lu>   This function does not support the ‘b’ directive, required by ISO C23,
    Po Lu>   on some platforms: glibc 2.34, musl libc, macOS 12.5, FreeBSD 13.1,
    Po Lu>   NetBSD 9.0, OpenBSD 7.2, AIX 7.2, Solaris 11.4, Cygwin 2.9.0.

    Po Lu> Glibc 2.37 does support this directive; Robert's Glibc is likely too
    Po Lu> old.

Only just, itʼs 2.36. We donʼt use 'b' anyway, but I guess this
issue will go away by itself.

Robert
--