gas64: changelog, tests

- gas64: fix casting ulong (with "negative value") to double/single (SARG)
- gas64: fix optimization prevented converting 32bit memory/register (SARG)

see c132cf3
see edee975

Author: jayrm

changelog.txt

@@ -95,6 +95,8 @@ Version 1.20.0
 - updated bindings: WinAPI: Allow FD_SETSIZE to be defined by user before including winsock.bi and winsock2.bi
 - sf.net #469, sf.net #721: check for unallocated dynamic arrays and wrong number of dimensions at run time and throw an error if attempting to access an unallocated array or indexing with the wrong number of dimensions
 - allow long ar and ld command lines in standalone windows builds (adeyblue)
+- gas64: fix casting ulong (with "negative value") to double/single (SARG)
+- gas64: fix optimization prevented converting 32bit memory/register (SARG)
 
 
 Version 1.10.3

tests/numbers/cast_f2l.bas

@@ -0,0 +1,100 @@
+#include "fbcunit.bi"
+
+'' show the compared numbers, but only if fbcu.getHideCases() is false
+#undef CU_ASSERT_EQUAL
+#define CU_ASSERT_EQUAL(a, b) if ((a) <> (b)) and (fbcu.getHideCases() = false) then print (a), (b) end if: CU_ASSERT( (a) = (b) )
+
+SUITE( fbc_tests.numbers.cast_f2l )
+
+	sub checkcast(byval x as double, byval ul as ulong)
+
+		'' 32 bit
+		CU_ASSERT_EQUAL( cuint( x ), cuint( ul ) )
+
+		if( ul > 1ul shl 31 ) then return
+		CU_ASSERT_EQUAL( cint( -x ), cint( -ul ) )
+		CU_ASSERT_EQUAL( clng( -x ), clng( -ul ) )
+
+		if( ul > (1ul shl 31) - 1 ) then return
+		CU_ASSERT_EQUAL( cint( x ), cint( ul ) )
+		CU_ASSERT_EQUAL( clng( x ), clng( ul ) )
+
+		if( ul > (1ul shl 31) - 1 ) then return
+		CU_ASSERT_EQUAL( cint( x ), cint( ul ) )
+		CU_ASSERT_EQUAL( clng( x ), clng( ul ) )
+
+		'' 16 bit
+		if( ul > 65535 ) then return
+		CU_ASSERT_EQUAL( cushort( x ), cushort( ul ) )
+
+		if( ul > 32768 ) then return
+		CU_ASSERT_EQUAL( cshort( -x ), cshort( -ul ) )
+
+		if( ul > 32767 ) then return
+		CU_ASSERT_EQUAL( cshort( x ), cshort( ul ) )
+
+		'' 8 bit
+		if( ul > 255 ) then return
+		CU_ASSERT_EQUAL( cubyte( x ), cubyte( ul ) )
+
+		if( ul > 128 ) then return
+		CU_ASSERT_EQUAL( cbyte( -x ), cbyte( -ul ) )
+
+		if( ul > 127 ) then return
+		CU_ASSERT_EQUAL( cbyte( x ), cbyte( ul ) )
+
+	end sub
+
+	sub testnum( byval n as ulong )
+
+		#define lsb(n) ((n) and -(n)) '' keep only least significant bit
+
+		'' only run when n has <= 53 significant bits
+		if( n and -(lsb(n) shl 53) ) then return
+
+		dim x as double = cdbl(n)
+
+		checkcast( x, n )
+
+		if( n < 1ul shl 31 ) then
+
+			'' make sure that cdbl(l) concurs with cdbl(ul)
+			CU_ASSERT_EQUAL( x, cdbl(clng(n)) )
+
+
+			if( n < 1ul shl 52 ) then
+				checkcast( x + 0.5, n + (n and 1) )
+				checkcast( x - 0.5, n - (n and 1) )
+			end if
+
+			if( n < 1ul shl 51 ) then
+				checkcast( x + 0.25, n )
+				checkcast( x - 0.25, n )
+			end if
+
+		end if
+
+	end sub
+
+	TEST( cast_l )
+
+		dim as long n = 1l
+		dim as double x = 1.0
+
+		dim as integer i, j, k, l
+
+		'' test powers of 2
+		for i = 0 to 31
+			testnum( 1ul shl i )
+		next i
+
+		'' test various bit combinations
+		for i = 0 to 31
+			for j = 0 to 31
+				testnum( (1 shl i) + (1 shl j) )
+			next j
+		next i
+
+	END_TEST
+
+END_SUITE

tests/numbers/cast_l2f.bas

@@ -0,0 +1,57 @@
+#include "fbcunit.bi"
+
+SUITE( fbc_tests.numbers.cast_l2f )
+
+	const c as longint = 1l shl 16
+
+	TEST( cast_ul )
+
+		dim as ulong n = 1ul
+		dim as double x = 1.0
+
+		for i as integer = 0 to 31
+
+			cu_assert_equal( cdbl(n), x )
+			if n <= not c then
+				cu_assert_equal( cdbl(n + c), x + c )
+			end if
+			if n >= c then
+				cu_assert_equal( cdbl(n - c), x - c )
+			end if
+
+			cu_assert_equal( cdbl(n + 1), x + 1.0 )
+			cu_assert_equal( cdbl(n - 1), x - 1.0 )
+
+			n shl= 1: x *= 2.0
+
+		next
+
+	END_TEST
+
+	TEST( cast_l )
+
+		dim as long n = 1l
+		dim as double x = 1.0
+
+		for i as integer = 0 to 30
+
+			cu_assert_equal( cdbl(n), x )
+			cu_assert_equal( cdbl(n + c), x + c )
+			cu_assert_equal( cdbl(n - c), x - c )
+
+			cu_assert_equal( cdbl(-n), -x )
+			cu_assert_equal( cdbl(-n + c), -x + c )
+			cu_assert_equal( cdbl(-n - c), -x - c )
+
+			cu_assert_equal( cdbl(n + 1), x + 1.0 )
+			cu_assert_equal( cdbl(n - 1), x - 1.0 )
+			cu_assert_equal( cdbl(-n + 1), -x + 1.0 )
+			cu_assert_equal( cdbl(-n - 1), -x - 1.0 )
+
+			n shl= 1: x *= 2.0
+
+		next
+                               
+	END_TEST
+
+END_SUITE