Re: [cygwin] cwrapper emits wrapper script

[Charles Wilson]

Ping?
http://lists.gnu.org/archive/html/libtool-patches/2007-04/msg00088.html

--
Chuck

Re: Fix `make dist' on systems with no F9x compiler

[Noah Misch]

On Fri, May 04, 2007 at 08:51:06AM +0200, Ralf Wildenhues wrote:
 * Noah Misch wrote on Fri, May 04, 2007 at 03:52:24AM CEST:
  `make dist' needs every test demo directory configured.  If the system has 
  no
  F90 compiler, tests/fcdemo/configure always fails.  This patch adds an 
  option to
  that configure script that suppresses the potentially fatal check.  We then 
  use
  the option when configuring a test demo directory for the sake of `make 
  dist'.
 
 Thanks for the patch.  Say, why not m4_pushdef([AC_MSG_ERROR],
 [something nonfatal]) ... m4_popdef([AC_MSG_ERROR]) instead?
 Just curious, I haven't tested or anything.

The fatal error is appropriate when fcdemo-conf.test configures this directory,
so we need a way to control it at `configure' time, not `autoconf' time.  There
certainly may be a solution more elegant than the one I proposed.

Rework a problematic use of ${1+$@}

[Noah Misch]

Zsh performs field splitting on `${1+$@}'; m4sh mostly insulates us from this
fact with `alias -g '${1+$@}'='$@''.  However, this alias is not used if the
parameter expansion is part of a larger word, such as `foo${1+$@}'.  This
patches func_echo to avoid the problem, fixing quote.test on Mac OS X 10.1.

2007-05-04  Noah Misch  [EMAIL PROTECTED]

* libltdl/config/general.m4sh (func_echo): Use $* instead of ${1+$@}.

diff -urpX dontdiff lt-bootstrap/libltdl/config/general.m4sh 
lt-echozsh/libltdl/config/general.m4sh
--- lt-bootstrap/libltdl/config/general.m4sh2007-03-25 07:12:42.0 
-0500
+++ lt-echozsh/libltdl/config/general.m4sh  2007-05-04 11:07:01.0 
-0400
@@ -129,7 +129,7 @@ opt_verbose=false
 # name if it has been set yet.
 func_echo ()
 {
-$ECHO $progname${mode+: }$mode: ${1+$@}
+$ECHO $progname${mode+: }$mode: $*
 }
 
 # func_verbose arg...

Skip dup convenience deplibs test more robustly

[Noah Misch]

powerpc-apple-dwarwin5.5 gets an unexpected pass for the duplicate convenience
deplibs test.  Rather than grow the list of xfail exclusions, I changed the test
to skip when the link command that might need a working --preserve-dup-deps
can succeed without it.

The test continues to xfail on GNU/Linux, but it now skips on Mac OS 10.1.  It
should now skip rather than pass on AIX.

2007-05-04  Noah Misch  [EMAIL PROTECTED]

* tests/duplicate_deps.at: Make the XFAIL unconditional, but skip the
test if the test link would succeed even without --preserve-dup-deps.

diff -urpX dontdiff lt-echozsh/tests/duplicate_deps.at 
lt-dupdepskip/tests/duplicate_deps.at
--- lt-echozsh/tests/duplicate_deps.at  2007-03-25 07:12:43.0 -0500
+++ lt-dupdepskip/tests/duplicate_deps.at   2007-05-04 11:44:35.0 
-0400
@@ -60,9 +60,15 @@ AT_CHECK([$LIBTOOL --mode=link --tag=CC 
  [0], [ignore], [ignore])
 LT_AT_EXEC_CHECK([./main])
 
-# This currently fails:
-AT_XFAIL_IF([case $host_os in aix*) false;; *) :;; esac])
+# Forgo the workaround and create a true circular dependency.
 $LIBTOOL --mode=link --tag=CC $CC $CFLAGS $LDFLAGS -o libb.la b1.lo
+# Some systems find all symbols anyway; skip this test on such systems.
+AT_CHECK([$LIBTOOL --mode=link --tag=CC \
+ $CC $CFLAGS $LDFLAGS -o main main.$OBJEXT liba.la libb.la liba.la \
+  exit 77],
+ [1], [ignore], [ignore])
+# Other systems need special treatment, which currently fails.
+AT_XFAIL_IF([:])
 AT_CHECK([$LIBTOOL --mode=link --preserve-dup-deps --tag=CC \
  $CC $CFLAGS $LDFLAGS -o main main.$OBJEXT liba.la libb.la liba.la],
  [0], [ignore], [ignore])