From f917da369a5a53c698eb5f063031862fd56b03b5 Mon Sep 17 00:00:00 2001
From: SteveMcNeill <smcneill@swva.net>
Date: Sat, 15 Aug 2015 10:48:35 -0400
Subject: [PATCH] Alteration to common.cpp to allow rounding to _UINT64_MAX
 limit

Floats previously only rounded to the upper limit of a signed integer64;
this alteration will now round them to the unsigned integer64 limit if
possible, or else toss an "OVERFLOW ERROR" if there's no way to squeeze
the float's value into an integer64.
---
 internal/c/common.cpp | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/internal/c/common.cpp b/internal/c/common.cpp
index ad90da999..340f411e8 100644
--- a/internal/c/common.cpp
+++ b/internal/c/common.cpp
@@ -228,6 +228,8 @@ struct img_struct{
 #define IMG_SCREEN 2 //img is linked to other screen pages
 #define IMG_FREEMEM 4 //if set, it means memory must be freed
 
+extern void error(int32 error_number); //declare the erorr handler so we can call it if needed.
+
 #ifdef QB64_NOT_X86
 inline int64 qbr(long double f){if (f<0) return(f-0.5f); else return(f+0.5f);}
 inline uint64 qbr_longdouble_to_uint64(long double f){if (f<0) return(f-0.5f); else return(f+0.5f);}
@@ -237,11 +239,15 @@ inline int32 qbr_double_to_long(double f){if (f<0) return(f-0.5f); else return(f
 //QBASIC compatible rounding via FPU:
 #ifdef QB64_MICROSOFT
 inline int64 qbr(long double f){
-  int64 i;
+  int64 i; int temp=0;
+  if (f<=9223372036854775808.5) {error(6); return 0;} // if the float is smaller than what an integer 64 could possible hold, toss an overflow error.
+  if (f>=18446744073709551615.5) {error(6); return 0;} // same result if the number is larger than what an integer 64 could possibly hold.
+  if (f>9223372036854775807) {temp=1;f=f-9223372036854775808;} //if it's too large for a signed int64, make it an unsigned int64 and return that value if possible.
   __asm{
-    fld   f
+	  fld   f
       fistp i
       }
+  if (temp) return i|0x8000000000000000;//+9223372036854775808;
   return i;
 }
 inline uint64 qbr_longdouble_to_uint64(long double f){
@@ -273,13 +279,17 @@ inline int32 qbr_double_to_long(double f){
 //FLDL=load double
 //FLDT=load long double
 inline int64 qbr(long double f){
-  int64 i;
+  int64 i; int temp=0;
+  if (f<=9223372036854775808.5) {error(6); return 0;} // if the float is smaller than what an integer 64 could possible hold, toss an overflow error.
+  if (f>=18446744073709551615.5) {error(6); return 0;} // same result if the number is larger than what an integer 64 could possibly hold.
+  if (f>9223372036854775807) {temp=1;f=f-9223372036854775808;} //if it's too large for a signed int64, make it an unsigned int64 and return that value if possible.
   __asm__ (
 	   "fldt %1;"
 	   "fistpll %0;"              
 	   :"=m" (i)
 	   :"m" (f)
 	   );
+  if (temp) return i|0x8000000000000000;// if it's an unsigned int64, manually set the bit flag
   return i;
 }
 inline uint64 qbr_longdouble_to_uint64(long double f){