Reorganize core library docs.

doc/site/core/num.markdown

@@ -0,0 +1,92 @@
+^title Num Class
+^category core
+
+**TODO**
+
+### **abs**
+
+The absolute value of the number.
+
+    :::dart
+    -123.abs // 123
+
+### **ceil**
+
+**TODO**
+
+### **cos**
+
+The cosine of the number.
+
+### **floor**
+
+**TODO**
+
+### **isNan**
+
+Whether the number is [not a number](http://en.wikipedia.org/wiki/NaN). This is
+`false` for normal number values and infinities, and `true` for the result of
+`0/0`, the square root of a negative number, etc.
+
+### **sin**
+
+The sine of the number.
+
+### **sqrt**
+
+The square root of the number. Returns `nan` if the number is negative.
+
+### **-** operator
+
+Negates the number.
+
+    :::dart
+    var a = 123
+    -a // -123
+
+### **-**(other), **+**(other), **/**(other), **\***(other) operators
+
+The usual arithmetic operators you know and love. All of them do 64-bit
+floating point arithmetic. It is a runtime error if the right-hand operand is
+not a number. Wren doesn't roll with implicit conversions.
+
+### **%**(denominator) operator
+
+The floating-point remainder of this number divided by `denominator`.
+
+It is a runtime error if `denominator` is not a number.
+
+### **<**(other), **>**(other), **<=**(other), **>=**(other) operators
+
+Compares this and `other`, returning `true` or `false` based on how the numbers
+are ordered. It is a runtime error if `other` is not a number.
+
+### **~** operator
+
+Performs *bitwise* negation on the number. The number is first converted to a
+32-bit unsigned value, which will truncate any floating point value. The bits
+of the result of that are then negated, yielding the result.
+
+### **&**(other) operator
+
+Performs bitwise and on the number. Both numbers are first converted to 32-bit
+unsigned values. The result is then a 32-bit unsigned number where each bit is
+`true` only where the corresponding bits of both inputs were `true`.
+
+It is a runtime error if `other` is not a number.
+
+### **|**(other) operator
+
+Performs bitwise or on the number. Both numbers are first converted to 32-bit
+unsigned values. The result is then a 32-bit unsigned number where each bit is
+`true` only where the corresponding bits of both inputs were `true`.
+
+It is a runtime error if `other` is not a number.
+
+### **..**(other) operator
+
+**TODO**
+
+### **...**(other) operator
+
+**TODO**