gawk-3.1.3-mbstate_t.patch for hpux

[Michael Elizabeth Chastain]

This patch improves multibyte support detection for gawk.  It's versus
gawk 3.1.3.  It supersedes my stupid little "#if defined(__hpux__)"
patch.

Some platforms, such as hpux 11.11, have multibyte functions such as
mbrtowc but do not have the mbstate_t type (this depends on the compiler
and compiler flags used).  Surprisingly, Single Unix Spec Version 3
appears to allow this behaviour: mbrtowc and related functions are
mandatory, but mbstate_t is available only if _XOPEN_SOURCE is defined
to enable XSI extensions.

There are comments in the readline 4.3 source that BeOS has a similar
profile: multibyte functions, but no mbstate_t type.

gawk really requires mbstate_t.  Look at strncasecmpmbs for example.  So
it's reasonable to add a HAVE_MBSTATE_T test.

After applying this patch, you have to touch (or regenerate) these
files:

  sleep 2 ; touch gawk-3.1.3/aclocal.m4
  sleep 2 ; touch gawk-3.1.3/configh.in
  sleep 2 ; touch gawk-3.1.3/configure
  sleep 2 ; touch gawk-3.1.3/patchlev.h
  sleep 2 ; touch gawk-3.1.3/version.c

If you don't get these timestamps right then the build process will try
to do it for you, but I had problems with that (and I'm not interested
in working on that part of the build machinery today).

Testing: i686-pc-linux-gnu, hppa-hp-hpux11.11, and ia64-hp-hpux11.23.
For the hpux platforms, I tested with:

  3 compilers:      aCC, ansi-c, gcc
  2 _XOPEN_SOURCE:  no _XOPEN_SOURCE, _XOPEN_SOURCE=500
  2 locales:        fr_FR.iso88591, fr_FR.utf8

The locale testing didn't make any difference because mbstate_t
wasn't defined in any of the configurations that successfully built
anyways.  After this patch, I can go after the _XOPEN_SOURCE problem.

I have asked for an FSF copyright assignment for gawk.  It's my
intention to donate my copyright for this work to the FSF.
(I already have copyright assignments for other FSF projects,
and I don't need an employer quitclaim).

Michael C

2003-12-10  Michael Chastain  <address@hidden>

        * README_d/README.hpux: Document configurations that work for me.
        * configure.ac: Add AC_TYPE_MBSTATE_T.
        * configure: Regenerate.
        * configh.in: Regenerate.
        * awk.h: Test HAVE_MBSTATE_T for multibyte support.
        * regex_internal.h: Likewise.
        * regex_internal.h: Include stdarg.h/varargs.h before wchar.h.

diff -u -r -N s0/README_d/README.hpux s1/README_d/README.hpux
--- s0/README_d/README.hpux     2002-06-20 10:42:46.000000000 -0400
+++ s1/README_d/README.hpux     2003-12-10 15:24:34.000000000 -0500
@@ -1,3 +1,21 @@
+2003-12-10 15:19:38 EST
+Michael Elizabeth Chastain <address@hidden>
+
+I built and tested gawk on hppa-hp-hpux11.11 and ia64-hp-hpux11.23.
+All the tests in the test suite passed.
+
+I built with these compilers:
+
+  gcc 3.3.2
+  hp ansi C from /opt/ansic/bin
+  hp aCC    from /opt/aCC/bin
+
+I ran into these problems:
+
+  NLS does not work; configure with --disable-nls.
+  -D_XOPEN_SOURCE=500 does not work.
+  Multibyte support is not available.
+
 Mon, 27 May 2002 17:55:46 +0800
 -------------------------------
 
diff -u -r -N s0/awk.h s1/awk.h
--- s0/awk.h    2003-06-29 08:52:40.000000000 -0400
+++ s1/awk.h    2003-12-10 12:14:26.000000000 -0500
@@ -73,7 +73,7 @@
 #ifdef HAVE_SIGNUM_H
 #include <signum.h>
 #endif
-#if defined(HAVE_MBRLEN) && defined(HAVE_MBRTOWC) && defined(HAVE_WCHAR_H) && 
defined(HAVE_WCTYPE_H)
+#if defined(HAVE_MBSTATE_T) && defined(HAVE_MBRLEN) && defined(HAVE_MBRTOWC) 
&& defined(HAVE_WCHAR_H) && defined(HAVE_WCTYPE_H)
 /* We can handle multibyte strings.  */
 #define MBS_SUPPORT
 #include <wchar.h>
diff -u -r -N s0/configh.in s1/configh.in
--- s0/configh.in       2003-07-04 13:49:45.000000000 -0400
+++ s1/configh.in       2003-12-10 14:37:01.000000000 -0500
@@ -129,6 +129,9 @@
 /* Define to 1 if you have the `mbrtowc' function. */
 #undef HAVE_MBRTOWC
 
+/* Define to 1 if <wchar.h> declares mbstate_t. */
+#undef HAVE_MBSTATE_T
+
 /* Define to 1 if you have the <mcheck.h> header file. */
 #undef HAVE_MCHECK_H
 
@@ -430,6 +433,9 @@
 /* Define to widest signed type if <inttypes.h> doesn't define. */
 #undef intmax_t
 
+/* Define to a type if <wchar.h> does not define. */
+#undef mbstate_t
+
 /* Define to `long' if <sys/types.h> does not define. */
 #undef off_t
 
diff -u -r -N s0/configure s1/configure
--- s0/configure        2003-07-04 13:49:09.000000000 -0400
+++ s1/configure        2003-12-10 14:37:01.000000000 -0500
@@ -9266,6 +9266,64 @@
 _ACEOF
 
 
+echo "$as_me:$LINENO: checking for mbstate_t" >&5
+echo $ECHO_N "checking for mbstate_t... $ECHO_C" >&6
+if test "${ac_cv_type_mbstate_t+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+#line $LINENO "configure"
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#          include <wchar.h>
+int
+main ()
+{
+mbstate_t x; return sizeof x;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
+  (eval $ac_compile) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+         { ac_try='test -s conftest.$ac_objext'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  ac_cv_type_mbstate_t=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+ac_cv_type_mbstate_t=no
+fi
+rm -f conftest.$ac_objext conftest.$ac_ext
+fi
+echo "$as_me:$LINENO: result: $ac_cv_type_mbstate_t" >&5
+echo "${ECHO_T}$ac_cv_type_mbstate_t" >&6
+   if test $ac_cv_type_mbstate_t = yes; then
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_MBSTATE_T 1
+_ACEOF
+
+   else
+
+cat >>confdefs.h <<\_ACEOF
+#define mbstate_t int
+_ACEOF
+
+   fi
 
   echo "$as_me:$LINENO: checking for long long" >&5
 echo $ECHO_N "checking for long long... $ECHO_C" >&6
diff -u -r -N s0/configure.ac s1/configure.ac
--- s0/configure.ac     2003-07-04 13:26:33.000000000 -0400
+++ s1/configure.ac     2003-12-10 12:14:26.000000000 -0500
@@ -159,6 +159,7 @@
 AC_TYPE_SIGNAL
 AC_SIZE_T
 AC_TYPE_GETGROUPS
+AC_TYPE_MBSTATE_T
 jm_AC_TYPE_LONG_LONG
 jm_AC_TYPE_UNSIGNED_LONG_LONG
 jm_AC_TYPE_INTMAX_T
diff -u -r -N s0/regex_internal.h s1/regex_internal.h
--- s0/regex_internal.h 2003-03-11 04:44:35.000000000 -0500
+++ s1/regex_internal.h 2003-12-10 13:49:35.000000000 -0500
@@ -39,6 +39,13 @@
 #include <stdlib.h>
 #include <string.h>
 
+/* HPUX 11.11 needs stdarg/varargs before wchar.h */
+#if defined(HAVE_STDARG_H) && defined(__STDC__) && __STDC__
+#include <stdarg.h>
+#else
+#include <varargs.h>
+#endif
+
 #if defined HAVE_LOCALE_H || defined _LIBC
 # include <locale.h>
 #endif
@@ -81,7 +88,7 @@
 # define gettext_noop(String) String
 #endif
 
-#if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && 
HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
+#if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && 
HAVE_MBSTATE_T && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
 # define RE_ENABLE_I18N
 #endif