[Advanced-java] Simple byte question

[Brian.Cain@xxxxxxxxxxxx (Cain Brian-BCAIN1)]

> -----Original Message-----
> From: Jason Boyd [mailto:jboyd@xxxxxxxxxxxxxxxx]
> 
> If (byte) 0x10 == 0001 0000
>    (byte) 0X20 == 0010 0000
>    (byte) 0x40 == 0100 0000
> 
> What is the equivalent of 1000 0000?

Is "1000 0000" in two's complement?

> Is it -0x01 ? -0xFF ?

I believe it's 0x80.  I've never tried the arithmetic negation operator with
hex, but I don't know if it's going to do what you want it to.  You should
try it to see what it does though.  Java does try to give you some heads up,
because when you try to assign a literal that sets the sign bit to a byte,
it forces you to cast it first (so that you're aware of things like sign
extension, I guess).

Since all data types are signed, you can do bitwise operations on a byte
that contains "0x80", and it will contain the leftmost bit set to one, all
others set to zero.

> Part 2: How does one use unsigned types within Java without 
> such trickery?

Dunno.  When using network protocols, I try to abstract things when I can
with DataIn/OutputStreams (which really do bitmasks and shifts, but
conveniently hide them from view).

Otherwise, you just have to deal with the fact that it's signed, I believe.
If I want 32 unsigned bits, I use a long (but only if it represents a
quantity).  If I need 32 bits for a unique identifier, then it doesn't
matter whether they're signed or not.

I think I considered using BigDecimal for a while, since it seemed to
acknowledge some things about sign, but that's usually like using a hammer
to crack a nut.  Arbitrary precision is not necessary.

If someone else knows a good workaround, I'm curious too.

-Brian