[Findutils-patches] [PATCH] Fix Savannah bug #20310 (sort -z works on host but not target)

[James Youngman]

2007-07-01  James Youngman  <address@hidden>

        Fix Savannah bug #20310.
        * m4/nullsort.m4 (jy_SORTZ): When cross-compiling, assume no
        support for sort -z.

Signed-off-by: James Youngman <address@hidden>
---
 NEWS           |    4 ++++
 m4/nullsort.m4 |   49 ++++++++++++++++++++++++++++++-------------------
 2 files changed, 34 insertions(+), 19 deletions(-)

diff --git a/NEWS b/NEWS
index aa74235..56b3fa3 100644
--- a/NEWS
+++ b/NEWS
@@ -7,6 +7,10 @@ Findutils version 4.3.9 is released under version 3 of the GNU 
General
 Public License.
 
 ** Bug Fixes
+#20310: configure uses hosts's support status for "sort -z" when
+generating athe updfatedb script for use on the target.  This is not a
+reliable strategy, so assume no support for -z when cross compiling.
+
 #20263: Compilation fix for DEC Alpha OSF/1 cc, which forbids the
 ordering comparison of function pointers.
 
diff --git a/m4/nullsort.m4 b/m4/nullsort.m4
index 8173448..e34627b 100644
--- a/m4/nullsort.m4
+++ b/m4/nullsort.m4
@@ -1,28 +1,39 @@
 AC_DEFUN([jy_SORTZ],
   [AC_PATH_PROG([SORT], [sort], [sort])
   AC_MSG_CHECKING([if $SORT supports the -z option])
-# find out if the sort command has a working -z option.
-if $SORT -z -c < "${srcdir:-.}/m4/order-good.bin" 2>/dev/null  >/dev/null
+if test "x$cross_compiling" = xyes
 then
-       # sort has a -z option, but we have not yet established that 
-       # sort thinks there is more than one input line there.   We have 
-       # to make sort -c do its thing with the input lines in the wrong 
-       # order to determine that (we can't do it in one shot because 
-       # if sort returns nonzero we cant tell that it wasn't just 
-       # complaining about this unknown -z option.
-       if $SORT -z -c < "${srcdir:-.}/m4/order-bad.bin" 2>/dev/null >/dev/null
-       then
-               # sort likes -z but it doesn't seem to make \0 
-               # a delimiter.
-               ac_sort_has_z=false
-       else
-               ac_sort_has_z=true
-       fi
+       # We are cross-compiling, so cannot test the target's "sort".
+       ac_sort_has_z=false
 else
-       # Doesn't like the z option.
-       ac_sort_has_z=false
+    # find out if the sort command has a working -z option.
+    if $SORT -z -c < "${srcdir:-.}/m4/order-good.bin" 2>/dev/null  >/dev/null
+    then
+       # sort has a -z option, but we have not yet established that 
+       # sort thinks there is more than one input line there.   We have 
+       # to make sort -c do its thing with the input lines in the wrong 
+       # order to determine that (we can't do it in one shot because 
+       # if sort returns nonzero we cant tell that it wasn't just 
+       # complaining about this unknown -z option.
+       if $SORT -z -c < "${srcdir:-.}/m4/order-bad.bin" 2>/dev/null >/dev/null
+       then
+               # sort likes -z but it doesn't seem to make \0 
+               # a delimiter.
+               ac_sort_has_z=false
+       else
+               ac_sort_has_z=true
+       fi
+    else
+       # Doesn't like the z option.
+       ac_sort_has_z=false
+    fi
 fi
-if $ac_sort_has_z
+
+if test "x$cross_compiling" = xyes
+then
+       AC_MSG_RESULT([assume no, since we are cross compiling])
+       AC_SUBST(SORT_SUPPORTS_Z,[false])
+elif $ac_sort_has_z
 then
        AC_MSG_RESULT([yes])
        AC_SUBST(SORT_SUPPORTS_Z,[true])
-- 
1.5.2.1