doubt in primitive casting..

[Nibin Jacob Panicker]

Hi all,

i have a doubt in implicit casting.
the following code doesnt compile and all of us know the reason for that.we need explicitely cast the result(b+1) to byte.
byte b = 1;
b = b + 1;

but what about
byte b = 1;
b += 1;
This piece of code compiles..howcome??does implicit casting happens in this case??

[Marcelo Ortega]

byte b = 1; //line 1
b = b + 1; //line 2

line 2 will evaluate to an integer (int) value of 32 bits, and knowing that a byte can only hold 8 bits, the compiler will flag an error.

b += 1;

In the preceding case, the compiler will add an implicit cast when this compound assignment operator is used.

[Nibin Jacob Panicker]

hi marzo..

so in the case of b+=1 implicit casting is happening so same should be happening in the case of b++ also.

but my doubt is if java can do implicit casting for b+=1 and b++ why not for b = b + 1; ??? anyway by giving a compile time error its just making us to do the job..

thanks

[Sakti Singh]

1. Java does implicit narrowing conversion when the right hand side of an expression is a constant or can be evaluated to be a constant during compile time.

so
byte b =1 //valid
byte b= 1+2 // valid

but byte b = b+1 // Invalid -- as it can not be evaluated during compile time

2. Above is valid only if the compile time constant is within the range of the type used.

byte b = 128; //Invalid -- compile time error

3. b+=1 is evaluated as b = (Type of b) b+1 by compiler.

So in the above case its b = (byte) b+1 which is correct since the integer value as a result of (b+1) is typecasted to byte.

Also note that :

byte b = 127;
b+=1; or b++ or ++b // Valid -- for the above reason


Hope this helps.

Regards,
Sakti
[ August 17, 2005: Message edited by: Sakti Singh ]

[Santana Iyer]

Sorry but b=(byte) b + 1; is wrong
it is b+=1; -> b = (byte) (b+1);

[Cecilia Jane]

Hi,

I just noticed..isn't byte b=(byte)b+1;

return an int? if it has to be byte then..it should be byte b=(byte)(b+1);


Cecilia

[Nibin Jacob Panicker]

thank you buddies..but my doubt was why cant java do implicit casting for
byte b=1;
b= b+1;

why cant java compiler take it as b = (byte)(b+1) during compile time??
if it does that during compile time there shouldnt be any harm whatever the value of b would be ..rgt??

[Cecilia Jane]

Hi,

What is the difference between compile time and run time actually... If a value of a variable is already known before it runs...then it is compile time..so when u say

byte b= 1;//this is ok...coz value of b is 1 which is a constant so you dont have to say byte b=(byte)1...that is at compile time the value of b is already decided before it runs..right.

now let is c the following statement...

b=b+1;

the value of b can vary...coz it is not a final variable..and taking this statement individually...when u say b= b+1..the value of b can vary it cud be anything...so it is not a constant meaning the value of b is deduced at run time..so it throws an error at compile time..

ok lets for a moment assume that complier implitcitly casts this...so there is gonna b loss or precision right? complier wud do it even if u dont know actually what happened..but when u say b=(byte)(b+1); u r actually saying to the compiler hey...I know what am doing..even though the value is changed am aware of it. Actually this makes the program lot easier.

Cecilia

[Sakti Singh]

my apologies , please read it as b = (byte)(b+1)

thanks to all those who pointed this out

Regards,
Sakti