add more docs

Author: sidorares

lib/packets/packet.js

@@ -374,16 +374,20 @@ Packet.prototype.readString = function (len) {
   return this.buffer.utf8Slice(this.offset - len, this.offset);
 };
 
-var minus = '-'.charCodeAt(0);
-var plus = '+'.charCodeAt(0);
-
 // The whole reason parse* function below exist
-// is because String creation is relatively expensive, and if we have
+// is because String creation is relatively expensive (at least with V8), and if we have
 // a buffer with "12345" content ideally we would like to bypass intermediate
 // "12345" string creation and directly build 12345 number out of
 // <Buffer 31 32 33 34 35> data.
-// In my benchmarks "parse" methods
+// In my benchmarks the difference is ~25M 8-digit numbers per second vs
+// 4.5 M using Number(packet.readLengthCodedString())
+// not used when size is close to max precision as series of *10 accumulate error
+// and approximate result mihgt be diffreent from (approximate as well) Number(bigNumStringValue))
+// In the futire node version if speed difference is smaller parse* functions might be removed
+// don't consider them as Packet public API
 
+var minus = '-'.charCodeAt(0);
+var plus = '+'.charCodeAt(0);
 
 Packet.prototype.parseInt = function (len, supportBigNumbers) {
 
@@ -475,7 +479,7 @@ Packet.prototype.parseIntNoBigCheck = function (len) {
     result += this.buffer[this.offset] - 48;
     this.offset++;
   }
-  return result * sign;
+  return  result * sign;
 };
 
 // copy-paste from https://github.com/felixge/node-mysql/blob/master/lib/protocol/Parser.js