hadoop - implement customized rawcomparator -

i need improve mr jobs, 1 thing think implement customized rawcomparator, key class have lots of fields string besides int fields, not sure how parse out string fields out of byte[],

my key class

public generalkey {   private int day;   private int hour;   private string type;   private string name;   .. } 

my customized rawcomparator:

public class generalkeycomparator extends writablecomparator {     private static final text.comparator text_comparator = new text.comparator();      protected generalkeycomparator() {         super(generalkey.class);     }      @override     public int compare(byte[] b1, int s1, int l1, byte[] b2, int s2, int l2) {         int day1 = readint(b1, s1);         int day2 = readint(b2, s2);          int comp = (intday1 < intday2) ? -1 : (intday1 == intday2) ? 0 : 1;         if (0 != comp) {             return comp;         }          int hr1 = readint(b1, s1+4);         int hr2 = readint(b2, s2+4);         comp = (hr1 < hr2) ? -1 : (hr1 == hr2) ? 0 : 1;              .... how compare string fields here???             return comp;     } 

google around found people tried :

try {     int firstl1 = writableutils.decodevintsize(b1[s1]) + readint(b1, s1+8);     int firstl2 = writableutils.decodevintsize(b2[s2]) + readvint(b2, s2+8);     comp = text_comparator.compare(b1, s1, firstl1, b2, s2, firstl2);  } catch (ioexception e) {     throw new illegalargumentexception(e); } 

but don't understand how work , don't think works in case, can help? thanks

added readfield() , write() methods here:

public void readfields(datainput input) throws ioexception {     intday = input.readint();     hr = input.readint();     type = input.readutf();     name = input.readutf();     ...     }  @override public void write(dataoutput output) throws ioexception {     output.writeint(intday);     output.writeint(hr);     output.writeutf(type);     output.writeutf(name);             ...     } 

you right. example found not work you. data fields in key example writablecomparables. have fundamental types (int, string) instead.

as using fundamental types, assume have implemented serialization / deserialization methods custom key type.

for third , fourth data field java strings, should able use compareto method on string class.

other option use writablecomparables instead of using fundamental types , use same technique found on google example.