What's the main difference between the two?
Still both of them are for writing Strings.
public void writeUTF(String str)
throws IOException
Primitive data write of this String in modified UTF-8 format.
vs
public void writeBytes(String str)
throws IOException
Writes a String as a sequence of bytes.
When should I use one rather than the other?
It's in the documentation... from DataOutput.writeBytes(String):
Writes a string to the output stream. For every character in the string s, taken in order, one byte is written to the output stream. If s is null, a NullPointerException is thrown.
If s.length is zero, then no bytes are written. Otherwise, the character s[0] is written first, then s1, and so on; the last character written is s[s.length-1]. For each character, one byte is written, the low-order byte, in exactly the manner of the writeByte method . The high-order eight bits of each character in the string are ignored.
In other words, "Sod Unicode, we don't care about any characters not in ISO-8859-1. Oh, and we assume you don't care about the length of the string either."
Note that writeBytes doesn't even try to detect data corruption - if you write out a character which isn't in ISO-8859-1, it will just drop the high byte silently.
Just say no - writeUTF is your friend... assuming your string is less than 64K in length.
Of course, if you have a protocol you're trying to implement which itself requires a single-byte encoding (ISO-8859-1 or ASCII) and doesn't use a length, then writeBytes might be appropriate - but I'd personally probably perform the text-to-bytes conversion myself and then use write(byte[]) instead... it's clearer.