You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@stdcxx.apache.org by vi...@apache.org on 2008/05/07 03:54:32 UTC

svn commit: r653965 - /stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp

Author: vitek
Date: Tue May  6 18:54:31 2008
New Revision: 653965

URL: http://svn.apache.org/viewvc?rev=653965&view=rev
Log:
2008-05-06  Travis Vitek  <vi...@roguewave.com>

	* tests/localization/22.locale.num.get.cpp: Revert r657262
	and apply a simpler fix.
	(test_flt_uflow): Make val static to avoid cadvise warning.
	(test_dbl_uflow): Ditto.
	(test_ldbl_uflow): Ditto.


Modified:
    stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp

Modified: stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp
URL: http://svn.apache.org/viewvc/stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp?rev=653965&r1=653964&r2=653965&view=diff
==============================================================================
--- stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp (original)
+++ stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp Tue May  6 18:54:31 2008
@@ -56,7 +56,6 @@
 
 /**************************************************************************/
 
-#include <cstddef>   // for std::size_t, std::ptrdiff_t
 #include <cfloat>    // FLT_MAX, FLT_MIN, etc.
 #include <climits>   // INT_MAX, INT_MIN, etc.
 #include <clocale>   // for localeconv(), setlocale()
@@ -222,7 +221,7 @@
                     _RWSTD_CONST_CAST (charT*, gend));
     }
 
-    std::ptrdiff_t gptr_off () const {
+    int gptr_off () const {
         return this->gptr () - this->eback ();
     }
 };
@@ -235,6 +234,13 @@
 
 /**************************************************************************/
 
+// if non-zero expected to point to a maximum valid value
+// of type T that's being tested below (used for floating
+// point ranges)
+// on function return, the pointed to value is overwritten
+// with the actual extracted value
+void *pmax = 0;
+
 
 enum IterType { iter_pointer, iter_istreambuf, iter_input };
 
@@ -244,7 +250,6 @@
              const char *cname,           // name of character type
              IterType    itype,           // type of input iterator
              const char *iname,           // name of input iterator
-             void       *pmax,
              nativeT     val,             // value expected to be extracted
              const char *str,             // input sequence
              int         eat_expect = -1, // number of consumed characters
@@ -315,7 +320,7 @@
 
     const charT *plast = pnext + std::char_traits<charT>::length (pnext);
 
-    std::ptrdiff_t consumed;
+    int consumed;
 
     switch (itype) {
     case iter_pointer: {
@@ -438,7 +443,6 @@
              const char *cname,           // name of character type
              IterType    itype,           // iterator type
              const char *iname,           // name of iterator type
-             void       *pmax,
              nativeT     val,             // value expected to be extracted
              const char *str,             // input sequence
              int         eat_expect = -1, // number of consumed characters
@@ -449,15 +453,15 @@
     switch (ctype) {
     case narrow_char:
         return do_test (lineno, (char*)0, cname, itype, iname,
-                        pmax, val, str, eat_expect, flags, err_expect, grouping);
+                        val, str, eat_expect, flags, err_expect, grouping);
     case wide_char:
         return do_test (lineno, (wchar_t*)0, cname, itype, iname,
-                        pmax, val, str, eat_expect, flags, err_expect, grouping);
+                        val, str, eat_expect, flags, err_expect, grouping);
     case user_char:
         break;
 #if 0   // disabled
         return do_test (lineno, (UserChar*)0, cname, itype, iname,
-                        pmax, val, str, eat_expect, flags, err_expect, grouping);
+                        val, str, eat_expect, flags, err_expect, grouping);
 #endif   // disabled
     }
 
@@ -497,8 +501,7 @@
 /*************************************************************************/
 
 #define TEST   do_test
-#define T      __LINE__, ctype, cname, itype, iname, 0
-#define T_MAX  __LINE__, ctype, cname, itype, iname, &val
+#define T      __LINE__, ctype, cname, itype, iname
 
 
 template <class numT>
@@ -2099,16 +2102,15 @@
     TEST (T, PVoid (0x0000ffff), "FfFf", 4, 0, Eof);
 
     // exercise overflow conditions
-#if    defined (_RWSTD_LONG_LONG) && _RWSTD_PTR_SIZE > _RWSTD_LONG_SIZE \
-    || ULONG_MAX > 0xffffffffUL
-    const char pvmax[]        = "0xffffffffffffffff";
-    const char pvmax_plus_1[] = "0x10000000000000000";
-#elif ULONG_MAX == 0xffffUL
+#if ULONG_MAX == 0xffffUL
     const char pvmax[]        = "0xffff";
     const char pvmax_plus_1[] = "0x10000";
 #elif ULONG_MAX == 0xffffffffUL
     const char pvmax[]        = "0xffffffff";
     const char pvmax_plus_1[] = "0x100000000";
+#elif ULONG_MAX > 0xffffffffUL
+    const char pvmax[]        = "0xffffffffffffffff";
+    const char pvmax_plus_1[] = "0x10000000000000000";
 #else
     // working around a SunPro bug (PR #28279)
     const char pvmax[]        = "0";
@@ -2194,7 +2196,7 @@
     TEST (T, F (1.0e+28), "10000000000000000000000000000",  29, 0, Eof);
     TEST (T, F (1.0e+29), "100000000000000000000000000000", 30, 0, Eof);
 
-#define VALSTR(x)   floatT (x), #x, int (std::strlen (#x))
+#define VALSTR(x)   floatT (x), #x, std::strlen (#x)
 
     // exercise various forms of floating point 0
 
@@ -2441,7 +2443,8 @@
              cname, iname);
 
     // exercise bahvior on underflow
-    float val = 0;
+    static float val = 0;
+    pmax = &val;
 
     // on underflow, get() follows C99 requirements on strtof()
     // i.e., it stores a value in the range [0, +/-FLT_MIN]
@@ -2452,34 +2455,35 @@
     //     normalized positive number in the return type; whether
     //     errno acquires the value ERANGE is implementation-defined.
 
-    TEST (T_MAX, (val = FLT_MIN, 0.0f), "1.111111e-9999", 14, 0, Eof);
+    TEST (T, (val = FLT_MIN, 0.0f), "1.111111e-9999", 14, 0, Eof);
     rw_assert (!(val < 0.0), 0, __LINE__,
                "correct sign after positive underflow");
 
-    TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.111111e-9999", 15, 0, Eof);
+    TEST (T, (val = 0.0f, -FLT_MIN), "-1.111111e-9999", 15, 0, Eof);
     rw_assert (!(val > 0.0), 0, __LINE__,
                "correct sign after negative underflow");
 
     if (1.234567e-39 < FLT_MIN) {
-        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-39", 12, 0, Eof);
-        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-39", 13, 0, Eof);
+        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-39", 12, 0, Eof);
+        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-39", 13, 0, Eof);
     }
 
     if (1.234567e-49 < FLT_MIN) {
-        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-49", 12, 0, Eof);
-        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-49", 13, 0, Eof);
+        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-49", 12, 0, Eof);
+        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-49", 13, 0, Eof);
     }
 
     if (1.234567e-99 < FLT_MIN) {
-        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-99", 12, 0, Eof);
-        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-99", 13, 0, Eof);
+        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-99", 12, 0, Eof);
+        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-99", 13, 0, Eof);
     }
 
     // exercise facet's behavior on underflow:
     //   parsing succeeds (fail is clear), +/-min is stored
-    TEST (T_MAX,  FLT_MIN, _RWSTD_STRSTR ( _RWSTD_DBL_MIN), -1, 0, Eof);
-    TEST (T_MAX, -FLT_MIN, _RWSTD_STRSTR (-_RWSTD_DBL_MIN), -1, 0, Eof);
+    TEST (T,  FLT_MIN, _RWSTD_STRSTR ( _RWSTD_DBL_MIN), -1, 0, Eof);
+    TEST (T, -FLT_MIN, _RWSTD_STRSTR (-_RWSTD_DBL_MIN), -1, 0, Eof);
 
+    pmax = 0;   // reset before next test
 
 
     rw_info (0, 0, 0, "std::num_get<%s, %s>::get (..., float&) on overflow",
@@ -2610,7 +2614,8 @@
              cname, iname);
 
     // exercise bahvior on underflow
-    double val = DBL_MIN;
+    static double val = DBL_MIN;
+    pmax = &val;
 
     // on underflow, get() follows C99 requirements on strtod()
     // i.e., it stores a value in the range [0, +/-DBL_MIN]
@@ -2621,24 +2626,25 @@
     //     normalized positive number in the return type; whether
     //     errno acquires the value ERANGE is implementation-defined.
 
-    TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.111111e-9999", 14, 0, Eof);
-    TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.111111e-9999", 15, 0, Eof);
+    TEST (T, (val = DBL_MIN, 0.0),   "1.111111e-9999", 14, 0, Eof);
+    TEST (T, (val = 0.0, -DBL_MIN), "-1.111111e-9999", 15, 0, Eof);
 
     if (1.23456789e-309L < DBL_MIN) {
-        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-309", 15, 0, Eof);
-        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-309", 16, 0, Eof);
+        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-309", 15, 0, Eof);
+        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-309", 16, 0, Eof);
     }
 
     if (1.234567e-409L < DBL_MIN) {
-        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-409", 15, 0, Eof);
-        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-409", 16, 0, Eof);
+        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-409", 15, 0, Eof);
+        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-409", 16, 0, Eof);
     }
 
     if (1.234567e-999L < DBL_MIN) {
-        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-999", 15, 0, Eof);
-        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-999", 16, 0, Eof);
+        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-999", 15, 0, Eof);
+        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-999", 16, 0, Eof);
     }
 
+    pmax = 0;   // reset before next test
 
 
 #  ifdef _RWSTD_LDBL_MAX
@@ -2731,7 +2737,8 @@
              cname, iname);
 
     // exercise bahvior on underflow
-    long double val = LDBL_MIN;
+    static long double val = LDBL_MIN;
+    pmax = &val;
 
     // on underflow, get() follows C99 requirements on strtold()
     // i.e., it stores a value in the range [0, +/-LDBL__MIN]
@@ -2742,22 +2749,23 @@
     //     normalized positive number in the return type; whether
     //     errno acquires the value ERANGE is implementation-defined.
 
-    TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "1.987654321e-99999", 18, 0, Eof);
-    TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-1.987654321e-99999", 19, 0, Eof);
+    TEST (T, (val = LDBL_MIN, 0.0L),   "1.987654321e-99999", 18, 0, Eof);
+    TEST (T, (val = 0.0L, -LDBL_MIN), "-1.987654321e-99999", 19, 0, Eof);
 
     const char fmt[] = "%Lg";
 
     long double ld = 0.0;
     if (1 == std::sscanf ("3.456789e-4932", fmt, &ld) && ld < LDBL_MIN) {
-        TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "3.456789e-4932", 14, 0, Eof);
-        TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-3.456789e-4932", 15, 0, Eof);
+        TEST (T, (val = LDBL_MIN, 0.0L),   "3.456789e-4932", 14, 0, Eof);
+        TEST (T, (val = 0.0L, -LDBL_MIN), "-3.456789e-4932", 15, 0, Eof);
     }
 
     if (1 == std::sscanf ("3.456789e-9999", fmt, &ld) && ld < LDBL_MIN) {
-        TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "3.456789e-9999", 14, 0, Eof);
-        TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-3.456789e-9999", 15, 0, Eof);
+        TEST (T, (val = LDBL_MIN, 0.0L),   "3.456789e-9999", 14, 0, Eof);
+        TEST (T, (val = 0.0L, -LDBL_MIN), "-3.456789e-9999", 15, 0, Eof);
     }
 
+    pmax = 0;   // reset before next test
 
 
 #  ifdef _RWSTD_LDBL_MAX



Re: svn commit: r653965 - /stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp

Posted by Martin Sebor <se...@roguewave.com>.
vitek@apache.org wrote:
> Author: vitek
> Date: Tue May  6 18:54:31 2008
> New Revision: 653965
> 
> URL: http://svn.apache.org/viewvc?rev=653965&view=rev
> Log:
> 2008-05-06  Travis Vitek  <vi...@roguewave.com>
> 
> 	* tests/localization/22.locale.num.get.cpp: Revert r657262
> 	and apply a simpler fix.

I think you might have reverted more than what you intended.
We probably want to keep this: 
http://svn.apache.org/viewvc?view=rev&revision=644435.

Martin

> 	(test_flt_uflow): Make val static to avoid cadvise warning.
> 	(test_dbl_uflow): Ditto.
> 	(test_ldbl_uflow): Ditto.
> 
> 
> Modified:
>     stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp
> 
> Modified: stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp
> URL: http://svn.apache.org/viewvc/stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp?rev=653965&r1=653964&r2=653965&view=diff
> ==============================================================================
> --- stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp (original)
> +++ stdcxx/branches/4.2.x/tests/localization/22.locale.num.get.cpp Tue May  6 18:54:31 2008
> @@ -56,7 +56,6 @@
>  
>  /**************************************************************************/
>  
> -#include <cstddef>   // for std::size_t, std::ptrdiff_t
>  #include <cfloat>    // FLT_MAX, FLT_MIN, etc.
>  #include <climits>   // INT_MAX, INT_MIN, etc.
>  #include <clocale>   // for localeconv(), setlocale()
> @@ -222,7 +221,7 @@
>                      _RWSTD_CONST_CAST (charT*, gend));
>      }
>  
> -    std::ptrdiff_t gptr_off () const {
> +    int gptr_off () const {
>          return this->gptr () - this->eback ();
>      }
>  };
> @@ -235,6 +234,13 @@
>  
>  /**************************************************************************/
>  
> +// if non-zero expected to point to a maximum valid value
> +// of type T that's being tested below (used for floating
> +// point ranges)
> +// on function return, the pointed to value is overwritten
> +// with the actual extracted value
> +void *pmax = 0;
> +
>  
>  enum IterType { iter_pointer, iter_istreambuf, iter_input };
>  
> @@ -244,7 +250,6 @@
>               const char *cname,           // name of character type
>               IterType    itype,           // type of input iterator
>               const char *iname,           // name of input iterator
> -             void       *pmax,
>               nativeT     val,             // value expected to be extracted
>               const char *str,             // input sequence
>               int         eat_expect = -1, // number of consumed characters
> @@ -315,7 +320,7 @@
>  
>      const charT *plast = pnext + std::char_traits<charT>::length (pnext);
>  
> -    std::ptrdiff_t consumed;
> +    int consumed;
>  
>      switch (itype) {
>      case iter_pointer: {
> @@ -438,7 +443,6 @@
>               const char *cname,           // name of character type
>               IterType    itype,           // iterator type
>               const char *iname,           // name of iterator type
> -             void       *pmax,
>               nativeT     val,             // value expected to be extracted
>               const char *str,             // input sequence
>               int         eat_expect = -1, // number of consumed characters
> @@ -449,15 +453,15 @@
>      switch (ctype) {
>      case narrow_char:
>          return do_test (lineno, (char*)0, cname, itype, iname,
> -                        pmax, val, str, eat_expect, flags, err_expect, grouping);
> +                        val, str, eat_expect, flags, err_expect, grouping);
>      case wide_char:
>          return do_test (lineno, (wchar_t*)0, cname, itype, iname,
> -                        pmax, val, str, eat_expect, flags, err_expect, grouping);
> +                        val, str, eat_expect, flags, err_expect, grouping);
>      case user_char:
>          break;
>  #if 0   // disabled
>          return do_test (lineno, (UserChar*)0, cname, itype, iname,
> -                        pmax, val, str, eat_expect, flags, err_expect, grouping);
> +                        val, str, eat_expect, flags, err_expect, grouping);
>  #endif   // disabled
>      }
>  
> @@ -497,8 +501,7 @@
>  /*************************************************************************/
>  
>  #define TEST   do_test
> -#define T      __LINE__, ctype, cname, itype, iname, 0
> -#define T_MAX  __LINE__, ctype, cname, itype, iname, &val
> +#define T      __LINE__, ctype, cname, itype, iname
>  
>  
>  template <class numT>
> @@ -2099,16 +2102,15 @@
>      TEST (T, PVoid (0x0000ffff), "FfFf", 4, 0, Eof);
>  
>      // exercise overflow conditions
> -#if    defined (_RWSTD_LONG_LONG) && _RWSTD_PTR_SIZE > _RWSTD_LONG_SIZE \
> -    || ULONG_MAX > 0xffffffffUL
> -    const char pvmax[]        = "0xffffffffffffffff";
> -    const char pvmax_plus_1[] = "0x10000000000000000";
> -#elif ULONG_MAX == 0xffffUL
> +#if ULONG_MAX == 0xffffUL
>      const char pvmax[]        = "0xffff";
>      const char pvmax_plus_1[] = "0x10000";
>  #elif ULONG_MAX == 0xffffffffUL
>      const char pvmax[]        = "0xffffffff";
>      const char pvmax_plus_1[] = "0x100000000";
> +#elif ULONG_MAX > 0xffffffffUL
> +    const char pvmax[]        = "0xffffffffffffffff";
> +    const char pvmax_plus_1[] = "0x10000000000000000";
>  #else
>      // working around a SunPro bug (PR #28279)
>      const char pvmax[]        = "0";
> @@ -2194,7 +2196,7 @@
>      TEST (T, F (1.0e+28), "10000000000000000000000000000",  29, 0, Eof);
>      TEST (T, F (1.0e+29), "100000000000000000000000000000", 30, 0, Eof);
>  
> -#define VALSTR(x)   floatT (x), #x, int (std::strlen (#x))
> +#define VALSTR(x)   floatT (x), #x, std::strlen (#x)
>  
>      // exercise various forms of floating point 0
>  
> @@ -2441,7 +2443,8 @@
>               cname, iname);
>  
>      // exercise bahvior on underflow
> -    float val = 0;
> +    static float val = 0;
> +    pmax = &val;
>  
>      // on underflow, get() follows C99 requirements on strtof()
>      // i.e., it stores a value in the range [0, +/-FLT_MIN]
> @@ -2452,34 +2455,35 @@
>      //     normalized positive number in the return type; whether
>      //     errno acquires the value ERANGE is implementation-defined.
>  
> -    TEST (T_MAX, (val = FLT_MIN, 0.0f), "1.111111e-9999", 14, 0, Eof);
> +    TEST (T, (val = FLT_MIN, 0.0f), "1.111111e-9999", 14, 0, Eof);
>      rw_assert (!(val < 0.0), 0, __LINE__,
>                 "correct sign after positive underflow");
>  
> -    TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.111111e-9999", 15, 0, Eof);
> +    TEST (T, (val = 0.0f, -FLT_MIN), "-1.111111e-9999", 15, 0, Eof);
>      rw_assert (!(val > 0.0), 0, __LINE__,
>                 "correct sign after negative underflow");
>  
>      if (1.234567e-39 < FLT_MIN) {
> -        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-39", 12, 0, Eof);
> -        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-39", 13, 0, Eof);
> +        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-39", 12, 0, Eof);
> +        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-39", 13, 0, Eof);
>      }
>  
>      if (1.234567e-49 < FLT_MIN) {
> -        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-49", 12, 0, Eof);
> -        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-49", 13, 0, Eof);
> +        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-49", 12, 0, Eof);
> +        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-49", 13, 0, Eof);
>      }
>  
>      if (1.234567e-99 < FLT_MIN) {
> -        TEST (T_MAX, (val = FLT_MIN, 0.0f),   "1.234567e-99", 12, 0, Eof);
> -        TEST (T_MAX, (val = 0.0f, -FLT_MIN), "-1.234567e-99", 13, 0, Eof);
> +        TEST (T, (val = FLT_MIN, 0.0f),   "1.234567e-99", 12, 0, Eof);
> +        TEST (T, (val = 0.0f, -FLT_MIN), "-1.234567e-99", 13, 0, Eof);
>      }
>  
>      // exercise facet's behavior on underflow:
>      //   parsing succeeds (fail is clear), +/-min is stored
> -    TEST (T_MAX,  FLT_MIN, _RWSTD_STRSTR ( _RWSTD_DBL_MIN), -1, 0, Eof);
> -    TEST (T_MAX, -FLT_MIN, _RWSTD_STRSTR (-_RWSTD_DBL_MIN), -1, 0, Eof);
> +    TEST (T,  FLT_MIN, _RWSTD_STRSTR ( _RWSTD_DBL_MIN), -1, 0, Eof);
> +    TEST (T, -FLT_MIN, _RWSTD_STRSTR (-_RWSTD_DBL_MIN), -1, 0, Eof);
>  
> +    pmax = 0;   // reset before next test
>  
>  
>      rw_info (0, 0, 0, "std::num_get<%s, %s>::get (..., float&) on overflow",
> @@ -2610,7 +2614,8 @@
>               cname, iname);
>  
>      // exercise bahvior on underflow
> -    double val = DBL_MIN;
> +    static double val = DBL_MIN;
> +    pmax = &val;
>  
>      // on underflow, get() follows C99 requirements on strtod()
>      // i.e., it stores a value in the range [0, +/-DBL_MIN]
> @@ -2621,24 +2626,25 @@
>      //     normalized positive number in the return type; whether
>      //     errno acquires the value ERANGE is implementation-defined.
>  
> -    TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.111111e-9999", 14, 0, Eof);
> -    TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.111111e-9999", 15, 0, Eof);
> +    TEST (T, (val = DBL_MIN, 0.0),   "1.111111e-9999", 14, 0, Eof);
> +    TEST (T, (val = 0.0, -DBL_MIN), "-1.111111e-9999", 15, 0, Eof);
>  
>      if (1.23456789e-309L < DBL_MIN) {
> -        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-309", 15, 0, Eof);
> -        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-309", 16, 0, Eof);
> +        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-309", 15, 0, Eof);
> +        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-309", 16, 0, Eof);
>      }
>  
>      if (1.234567e-409L < DBL_MIN) {
> -        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-409", 15, 0, Eof);
> -        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-409", 16, 0, Eof);
> +        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-409", 15, 0, Eof);
> +        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-409", 16, 0, Eof);
>      }
>  
>      if (1.234567e-999L < DBL_MIN) {
> -        TEST (T_MAX, (val = DBL_MIN, 0.0),   "1.23456789e-999", 15, 0, Eof);
> -        TEST (T_MAX, (val = 0.0, -DBL_MIN), "-1.23456789e-999", 16, 0, Eof);
> +        TEST (T, (val = DBL_MIN, 0.0),   "1.23456789e-999", 15, 0, Eof);
> +        TEST (T, (val = 0.0, -DBL_MIN), "-1.23456789e-999", 16, 0, Eof);
>      }
>  
> +    pmax = 0;   // reset before next test
>  
>  
>  #  ifdef _RWSTD_LDBL_MAX
> @@ -2731,7 +2737,8 @@
>               cname, iname);
>  
>      // exercise bahvior on underflow
> -    long double val = LDBL_MIN;
> +    static long double val = LDBL_MIN;
> +    pmax = &val;
>  
>      // on underflow, get() follows C99 requirements on strtold()
>      // i.e., it stores a value in the range [0, +/-LDBL__MIN]
> @@ -2742,22 +2749,23 @@
>      //     normalized positive number in the return type; whether
>      //     errno acquires the value ERANGE is implementation-defined.
>  
> -    TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "1.987654321e-99999", 18, 0, Eof);
> -    TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-1.987654321e-99999", 19, 0, Eof);
> +    TEST (T, (val = LDBL_MIN, 0.0L),   "1.987654321e-99999", 18, 0, Eof);
> +    TEST (T, (val = 0.0L, -LDBL_MIN), "-1.987654321e-99999", 19, 0, Eof);
>  
>      const char fmt[] = "%Lg";
>  
>      long double ld = 0.0;
>      if (1 == std::sscanf ("3.456789e-4932", fmt, &ld) && ld < LDBL_MIN) {
> -        TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "3.456789e-4932", 14, 0, Eof);
> -        TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-3.456789e-4932", 15, 0, Eof);
> +        TEST (T, (val = LDBL_MIN, 0.0L),   "3.456789e-4932", 14, 0, Eof);
> +        TEST (T, (val = 0.0L, -LDBL_MIN), "-3.456789e-4932", 15, 0, Eof);
>      }
>  
>      if (1 == std::sscanf ("3.456789e-9999", fmt, &ld) && ld < LDBL_MIN) {
> -        TEST (T_MAX, (val = LDBL_MIN, 0.0L),   "3.456789e-9999", 14, 0, Eof);
> -        TEST (T_MAX, (val = 0.0L, -LDBL_MIN), "-3.456789e-9999", 15, 0, Eof);
> +        TEST (T, (val = LDBL_MIN, 0.0L),   "3.456789e-9999", 14, 0, Eof);
> +        TEST (T, (val = 0.0L, -LDBL_MIN), "-3.456789e-9999", 15, 0, Eof);
>      }
>  
> +    pmax = 0;   // reset before next test
>  
>  
>  #  ifdef _RWSTD_LDBL_MAX
> 
>