? about JAVA mathematics

[MrDingleDangle]

i am using nubers in the Short class and want to do math operations, but whenever I try to do them i recieve an error "possible loss of precision"...

 

[DaveSimmons]

are you mixing types, or using non-integer values anywhere? For example
foo = bar / 2.5
even if foo, bar are both short, bar has to be converted to float or double for the division, then converted back to short with, surprise, possible loss of precision.

 

[wfay]

pls post sample code with comments showing where it breaks, and sample inputs, and we can provide better assistance...

 

[damiano]

start using native double.... haha

why are you using short affraid of running out of memory?

The whole thing is that short uses less memory, so the precision is less.
think about it...
and unless you machine has 5 megs of ram or your system has a LOT of transactions... just use double

 

[MrDingleDangle]

im using short because it is what is returned by a program not written by me.

i am trying to subtract two shorts or multimply a short by an int...as follows

temp[x][y] = 4*temparray[x][y]-temparray[x+1][y]-temparray[x-1][y]-temparray[x][y+1]-temparray[x][y-1];

temp and temparray are short[][]

 

[damiano]

cast the return value to a double or a float...

once again haha

 

[singh]

Why does the compiler report an "error" instead of a warning? Anyone care to postulate?

 

[manly]

Basically, numeric types get promoted to a wider type when you do mixed-type arithmetic. While the rules involved are not trivial, in general you can simply assume numeric math is performed with ints and floating point math with doubles.

So if you're manipulating ints and want to store the result into a short, then you'll have to cast (narrow) the type to short.

 

[Rahminator]

Originally posted by: singh
Why does the compiler report an "error" instead of a warning? Anyone care to postulate?

Because Java blows chunks.

 

[MrDingleDangle]

Originally posted by: manly
Basically, numeric types get promoted to a wider type when you do mixed-type arithmetic. While the rules involved are not trivial, in general you can simply assume numeric math is performed with ints and floating point math with doubles.

So if you're manipulating ints and want to store the result into a short, then you'll have to cast (narrow) the type to short.

thanks thats getting me on the right track...for some reason i still get the error when i put (short) at the start of the operation...just keeps telling me i might be losing precision...and im just sitting here yelling at the screen that i know and that i dont mind...but that doesnt help

 

[manly]

Just to be clear, did you do did exactly this:

temp[x][y] = (short) (4*temparray[x][y]-temparray[x+1][y]-temparray[x-1][y]-temparray[x][y+1]-temparray[x][y-1]);

A narrowing cast might issue a compiler warning, but it doesn't raise an error. Also, I don't see why you can't be using int or int[][] even though you feel otherwise.

 

[MrDingleDangle]

Originally posted by: manly
Just to be clear, did you do did exactly this:

temp[x][y] = (short) (4*temparray[x][y]-temparray[x+1][y]-temparray[x-1][y]-temparray[x][y+1]-temparray[x][y-1]);

A narrowing cast might issue a compiler warning, but it doesn't raise an error. Also, I don't see why you can't be using int or int[][] even though you feel otherwise.

i hate programming...didnt have the () around the statement..just the short...

thanx!

 

[Ameesh]

Originally posted by: MrDingleDangle

Originally posted by: manly
Just to be clear, did you do did exactly this:

temp[x][y] = (short) (4*temparray[x][y]-temparray[x+1][y]-temparray[x-1][y]-temparray[x][y+1]-temparray[x][y-1]);

A narrowing cast might issue a compiler warning, but it doesn't raise an error. Also, I don't see why you can't be using int or int[][] even though you feel otherwise.

i hate programming...didnt have the () around the statement..just the short...

thanx!

when you put the the parens around it its called a type cast, you are basically telling the compiler this is the type of the variable coming next, if you dont have the parens you are beggining a declaration.

 

[manly]

Originally posted by: Ameesh

when you put the the parens around it its called a type cast, you are basically telling the compiler this is the type of the variable coming next, if you dont have the parens you are beggining a declaration.

I'm pretty sure he means that he didn't put parens around the expression, thus the (short) cast applied only to the first term of the expression (namely 4), instead of to the evaluation of the whole expression.