/usr/bin/nm and dlopen fix for HP-UX 10.20, 11.00

[libtool-patches]

/usr/bin/nm -p on HP-UX outputs one or two symbols. The current case
catches only one symbol. The patch below fixes this so two symbols are
recognized. The correct fix is to propagate $symcode to ltmain.sh
somehow but I've taken the short approach. I have run 'make check' and
all tests pass. This patch is against the 1.4 branch.

I'd also like to recommend a policy decision change in libtool.m4. We
should check for shl_load *first* on HP-UX. While dlopen() is
available, it works only when a patch is applied. Therefore, I suggest
we default to shl_load on HP-UX which requires no patch to work
properly.

2001-06-20  Albert Chin-A-Young  <address@hidden>

        * libtool.m4: Quote $symcode. Prefer shl_load to dlopen
          on HP-UX because dlopen won't work properly without
          a patch.

        * ltmain.in: Handle case when output of /usr/bin/nm -p
          emits more than one symbol. HP-UX sometimes emits two
          symbols. Correct thing to do is recognized as many
          $symcode symbols as output but we don't have access
          to $symcode here.

-- 
albert chin (address@hidden)

-- snip snip
Index: libtool.m4
===================================================================
RCS file: /home/cvs/libtool/libtool.m4,v
retrieving revision 1.166
diff -u -3 -p -r1.166 libtool.m4
--- libtool.m4  2001/04/24 23:40:42     1.166
+++ libtool.m4  2001/06/20 18:32:06
@@ -188,7 +188,7 @@ cygwin* | mingw* | pw32*)
   [symcode='[ABCDGISTW]']
   ;;
 hpux*) # Its linker distinguishes data from code symbols
-  lt_cv_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern char 
\1();/p' -e 's/^$symcode* .* \(.*\)$/extern char \1;/p'"
+  lt_cv_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern char 
\1();/p' -e 's/^'$symcode'* .* \(.*\)$/extern char \1;/p'"
   ;;
 irix*)
   [symcode='[BCDEGRST]']
@@ -614,13 +614,14 @@ else
    ;;
 
   *)
-    AC_CHECK_LIB(dl, dlopen,  [lt_cv_dlopen="dlopen" lt_cv_dlopen_libs="-ldl"],
-      [AC_CHECK_FUNC(dlopen, lt_cv_dlopen="dlopen",
-        [AC_CHECK_FUNC(shl_load, lt_cv_dlopen="shl_load",
-          [AC_CHECK_LIB(svld, dlopen,
-           [lt_cv_dlopen="dlopen" lt_cv_dlopen_libs="-lsvld"],
-            [AC_CHECK_LIB(dld, shl_load,
-              [lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-dld"])
+    AC_CHECK_FUNC(shl_load, lt_cv_dlopen="shl_load",
+      [AC_CHECK_LIB(dld, shl_load,
+        [lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-dld"],
+        [AC_CHECK_LIB(dl, dlopen,
+          [lt_cv_dlopen="dlopen" lt_cv_dlopen_libs="-ldl"],
+          [AC_CHECK_FUNC(dlopen, lt_cv_dlopen="dlopen",
+            [AC_CHECK_LIB(svld, dlopen,
+             [lt_cv_dlopen="dlopen" lt_cv_dlopen_libs="-lsvld"])
            ])
           ])
         ])
Index: ltmain.in
===================================================================
RCS file: /home/cvs/libtool/ltmain.in,v
retrieving revision 1.259
diff -u -3 -p -r1.259 ltmain.in
--- ltmain.in   2001/04/24 23:22:34     1.259
+++ ltmain.in   2001/06/20 18:32:07
@@ -3302,8 +3306,13 @@ lt_preloaded_symbols[] =
 {\
 "
 
+           # Ick! No guarantee that symbol is one character. On
+           # HP-UX, it could be two. But what happens when some
+           # other OS has three or more? We no longer have
+           # access to $symcode though.
            sed -n -e 's/^: \([^ ]*\) $/  {\"\1\", (lt_ptr_t) 0},/p' \
                -e 's/^. \([^ ]*\) \([^ ]*\)$/  {"\2", (lt_ptr_t) \&\2},/p' \
+               -e 's/^.. \([^ ]*\) \([^ ]*\)$/  {"\2", (lt_ptr_t) \&\2},/p' \
                  < "$nlist" >> "$output_objdir/$dlsyms"
 
            $echo >> "$output_objdir/$dlsyms" "\