patch 8.2.5004: right shift on negative number does not work as documented
Problem: Right shift on negative number does not work as documented.
Solution: Use a uvarnumber_T type cast.
diff --git a/runtime/doc/eval.txt b/runtime/doc/eval.txt
index f9443fd..77822d7 100644
--- a/runtime/doc/eval.txt
+++ b/runtime/doc/eval.txt
@@ -1138,9 +1138,10 @@
*E1282* *E1283*
The "<<" and ">>" operators can be used to perform bitwise left or right shift
of the left operand by the number of bits specified by the right operand. The
-operands must be positive numbers. The topmost bit (sign bit) is always
-cleared for ">>". If the right operand (shift amount) is more than the
-maximum number of bits in a number (|v:numbersize|) the result is zero.
+operands are used as positive numbers. When shifting right with ">>" the
+topmost bit (somtimes called the sign bit) is cleared. If the right operand
+(shift amount) is more than the maximum number of bits in a number
+(|v:numbersize|) the result is zero.
expr6 and expr7 *expr6* *expr7* *E1036* *E1051*
@@ -1417,6 +1418,10 @@
Decimal, Hexadecimal (starting with 0x or 0X), Binary (starting with 0b or 0B)
and Octal (starting with 0, 0o or 0O).
+Assuming 64 bit numbers are used (see |v:numbersize|) an unsigned number is
+truncated to 0x7fffffffffffffff or 9223372036854775807. You can use -1 to get
+0xffffffffffffffff.
+
*floating-point-format*
Floating point numbers can be written in two forms: