[FYI] Libobj tests: refer to automake bug tracker for limitations.

[Stefano Lattarini]

* tests/libobj15b.test (configure.in): In the comment explaining
why an apparently redundant AC_OUTPUT is indeed needed, refer
to the relevant entry in the Automake bug tracker rather than
describing the bug in detail with a FIXME-style comment.
* tests/libobj15c.test: Likewise.
* tests/libobj20b.test: Likewise.
---
 ChangeLog            |    8 ++++++++
 tests/libobj15b.test |    5 ++---
 tests/libobj15c.test |    5 ++---
 tests/libobj20b.test |    5 ++---
 4 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index b98aa53..3b9c1c4 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,13 @@
 2010-12-13  Stefano Lattarini  <address@hidden>
 
+       Libobj tests: refer to automake bug tracker for limitations.
+       * tests/libobj15b.test (configure.in): In the comment explaining
+       why an apparently redundant AC_OUTPUT is indeed needed, refer
+       to the relevant entry in the Automake bug tracker rather than
+       describing the bug in detail with a FIXME-style comment.
+       * tests/libobj15c.test: Likewise.
+       * tests/libobj20b.test: Likewise.
+
        Tests: do not use `##' automake comments in-line.
        * tests/libobj19.test (Makefile.am): Do not use inline `##'
        comments in the definition of AUTOMAKE_OPTIONS.
diff --git a/tests/libobj15b.test b/tests/libobj15b.test
index 000ba1f..3816602 100755
--- a/tests/libobj15b.test
+++ b/tests/libobj15b.test
@@ -24,9 +24,8 @@ cat >> configure.in << 'END'
 AC_PROG_CC
 AC_PROG_RANLIB
 AC_LIBSOURCE([foobar.c])
-# FIXME: without a call to AC_OUTPUT, Automake does not fail when file(s)
-#        specified in AC_LIBSOURCE call(s) do not exist.  Should we fix
-#        this limitation?  Or is the current behaviour good enough?
+# NOTE: this call to AC_OUTPUT is really needed; see Automake bug #7635
+#       <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7635>
 AC_OUTPUT
 END
 
diff --git a/tests/libobj15c.test b/tests/libobj15c.test
index f1aa259..29e0aa7 100755
--- a/tests/libobj15c.test
+++ b/tests/libobj15c.test
@@ -24,9 +24,8 @@ cat >> configure.in << 'END'
 AC_PROG_CC
 AC_PROG_RANLIB
 AC_LIBSOURCES([foobar.c, bazquux.c])
-# FIXME: without a call to AC_OUTPUT, Automake does not fail when file(s)
-#        specified in AC_LIBSOURCES call(s) do not exist.  Should we fix
-#        this limitation?  Or is the current behaviour good enough?
+# NOTE: this call to AC_OUTPUT is really needed; see Automake bug #7635
+#       <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7635>
 AC_OUTPUT
 END
 
diff --git a/tests/libobj20b.test b/tests/libobj20b.test
index 1662110..f57a78b 100755
--- a/tests/libobj20b.test
+++ b/tests/libobj20b.test
@@ -26,9 +26,8 @@ AC_CONFIG_LIBOBJ_DIR([libobj-dir])
 AC_PROG_CC
 AC_PROG_RANLIB
 AC_LIBSOURCE([foobar.c])
-# FIXME: without a call to AC_OUTPUT, Automake does not fail when file(s)
-#        specified in AC_LIBSOURCE call(s) do not exist.  Should we fix
-#        this limitation?  Or is the current behaviour good enough?
+# NOTE: this call to AC_OUTPUT is really needed; see Automake bug #7635
+#       <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7635>
 AC_OUTPUT
 END
 
-- 
1.7.1