Question

I’m working with eclipse IDE (Version: 3.4.2) on a mac and I have met the following issue.

When comparing between strings using equal() or equalsIgnoreCase() methods I receive false even when the string are equal. For example, the code below consider the following condition as false, even when values[0] = “debug_mode”

if (values[0].equalsIgnoreCase("debug_mode")) 
    debug_mode = true;

which is part of the following loop:

String value = dis.readLine();
String values[] = value.trim().split("=");
if (values.length >= 2)
{
    Config.prnt_dbg_msg(values[0] + "\t" + values[1]);
    if (values[0].equalsIgnoreCase("debug_mode")) 
        debug_mode = isTrue(values[1]);
    if (values[0].equalsIgnoreCase("debug_query_parsing")) 
        debug_query_parsing = isTrue(values[1]);
    if (values[0].equalsIgnoreCase("username")) 
        Connection_Manager.alterAccessParameters(values[1], null, null);
    if (values[0].equalsIgnoreCase("password")) 
        Connection_Manager.alterAccessParameters(null, values[1], null);
if (values[0].equalsIgnoreCase("database")) 
        Connection_Manager.alterAccessParameters(null, null, values[1]);
    if (values[0].equalsIgnoreCase("allow_duplicate_entries")) 
        allow_duplicate_entries = isTrue(values[1]);
}                         

I tried to use value[0].equal("debug_mode") and got the same result.
Does someone have any idea why?

Answer 1

That would be very strange indeed 🙂 Can you change the above code to this:

if ("debug_mode".equalsIgnoreCase("debug_mode")) 
    debug_mode = true;

confirm it works fine and then double check why your values[0] is not “debug_mode”.

Here’s what comes to my mind right now as a list of things to check:

• Check that values[0].length() == "debug_mode".length()
• I highly doubt, but let me put it on the table anyway – are you by any chance using Unicode?
• Can you print each character and do .equals() between that character and the respective character of the “debug_mode” string?
• If this is in a bigger project, can you do the same in a simple Java project and confirm it works there?

To clarify, the problem is actually using DataInputStream.readLine. From javadoc (http://download.oracle.com/javase/1.6.0/docs/api/java/io/DataInputStream.html):

readLine()
      Deprecated. This method does not properly convert bytes to characters. ...

It actually has to do with Unicode in a subtle way – when you do writeChar you actually write two bytes 0 and 97, big-endian Unicode for the letter a.

Here’s a self-contained snippet that shows the behavior:

import java.io.*;
import java.util.*;

public class B {
  public static void main(String[] args) throws Exception {
    String os = "abc";

    System.out.println("---- unicode, big-endian");
    for(byte b: os.getBytes("UTF-16BE")) {
      System.out.println(b);
    }

    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    DataOutputStream dos = new DataOutputStream(baos);

    for(char c: os.toCharArray()) {
      dos.writeChar(c);
    }

    byte[] ba = baos.toByteArray();

    System.out.println("---- ba");
    for(byte b: ba) {
      System.out.println(b);
    }

    ByteArrayInputStream bais = new ByteArrayInputStream(ba);
    DataInputStream dis = new DataInputStream(bais);

    System.out.println("---- dis");
    String s = dis.readLine();
    System.out.println(s);
    System.out.println("String length is " + s.length() 
      + ", but you would expect " + os.length() 
      + ", as that is what you see printed...");
  }
}

Moral of the story – don’t use deprecated api… Also, whitespace is the silent killer: http://www.codinghorror.com/blog/2009/11/whitespace-the-silent-killer.html