[markdb]

Hi,
forgive me if I'm being stupid or this
has been asked before :

Java only has signed bytes, which store
a maximum value of +127.

The Nokia MIDI to OTT converter outputs
a series of bytes, which are often above
127 (7F in hex). (Eg D1 and
95, which always appear in the OTT
header.)

One of the converter's output choices is
the code for a byte array, with these invalid values in it - this isn't valid Java!
The compiler doesn't complain about the
array definition, but try manually
assigning one of the high values from the
array to a byte variable later in the code;
You'll get a "loss of precision" error.

Please set me straight here - why does the converter produce byte values which java can't represent?

[walkal]

I think the explanation is that a byte with the leading bit set to 1 is a valid representation of a negative number, coded according to the twos-complement scheme. You can set a byte variable to such a value by casting an integer or a short to a byte. E.g.

byte b = (byte)0xd1;

The cast just removes the leading zeroes. If you then do a println on b, you will see it interpreted as a negative number (-47 in this case).

If you have a tone sequence stored as a string - e.g. 024a3a40... - you can convert it to bytes by doing something like this:

for (int i = 0; i < toneString.length(); i += 2) {
oneByte = toneString.substring(i, i + 2);
outStream.writeByte((byte)Short.parseShort(oneByte, 16));
}

There may well be a less cumbersome way of doing this, but this is one method that works.


Alan Walker