[PATCH] update automake version to 1.11.6

[Cary Coutant]

> I personally do verify the changes in the configure files,
> for instance, and ask myself whether each hunk I see makes sense
> to me or not. Seeing unrelated changes because others used a different
> version makes that process a little harder (and, most of the time,
> I'll just start over, and push a patch that first regenerates the
> file).

I have two machines, both with automake-1.11.1 configured and built
from the same source, but the two generate ever-so-slightly different
gold/Makefile.in:

--- a/gold/Makefile.in
+++ b/gold/Makefile.in
@@ -70,8 +70,8 @@ subdir = .
 DIST_COMMON = NEWS README ChangeLog $(srcdir)/Makefile.in \
        $(srcdir)/Makefile.am $(top_srcdir)/configure \
        $(am__configure_deps) $(srcdir)/config.in \
-       $(srcdir)/../mkinstalldirs $(top_srcdir)/po/Make-in ffsll.c \
-       ftruncate.c pread.c mremap.c yyscript.h yyscript.c \
+       $(srcdir)/../mkinstalldirs $(top_srcdir)/po/Make-in pread.c \
+       ffsll.c mremap.c ftruncate.c yyscript.h yyscript.c \
        $(srcdir)/../depcomp $(srcdir)/../ylwrap
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/../config/depstand.m4 \

I have to filter out this difference every time I run automake at home.

Both generate identical results for gold/testsuite/Makefile.in, though.

Is there a known non-determinism somewhere in automake? Maybe 1.11.6 fixed that.

-cary