Looping over arrays.

[Marc Twain]

Isn't it a waste of resources to loop over arrays this way:

for (i = 0; i < myArray.length; i ++) {
// do something
}

.... as opposed to create a variable to hold the value of
myArray.length first, like so:

int len = myArray.length;
for (i = 0; i < len; i ++) {
// do something
}

This way the length of the array doesn't have to be recalculated on
every iteration? Or am I completely wrong?

TIA,

AJ

 

[Christophe Vanfleteren]

Marc Twain

Isn't it a waste of resources to loop over arrays this way:

for (i = 0; i < myArray.length; i ++) {
// do something
}

... as opposed to create a variable to hold the value of
myArray.length first, like so:

int len = myArray.length;
for (i = 0; i < len; i ++) {
// do something
}

This way the length of the array doesn't have to be recalculated on
every iteration? Or am I completely wrong?

TIA,

AJ

The length of the array is not recalculated every time, since the length of
an array is known the moment it is created.
It wouldn't even be possible to recalculate, since length is a public int
field, and not a method.

So I would definitely go for the first version, since it is clearer to read.
The second version is good if the calculation of the length is an expensive
operation.

 

[Chris Uppal]

The length of the array is not recalculated every time, since the length
of an array is known the moment it is created.
It wouldn't even be possible to recalculate, since length is a public int
field, and not a method.

Actually, it's not a field either -- it's a (very very) nasty hack in the
compiler that fakes arrays having a field named "length". In compiling for the
JVM it gets translated into a special JVM bytecode that retrieves the size of
the array. Of course that bytecode is then subject to whatever kind of
optimisation the JVM does -- for instance the JIT (if there is one) could
replace the code with a variable.

For the OP: it is highly unlikely that it'd make much difference whichever way
you expressed it. Go with the one that's easiest to read. (There's nothing
wrong with considering such performance issues as a way of learning and
deepening your understanding, but when it comes to production code the general
advise about such optimisations is "don't".)

-- chris

 

[Marc Twain]

So I would definitely go for the first version, since it is clearer to read.

The second version is good if the calculation of the length is an expensive
operation.

Thanks!

AJ

 

[Roedy Green]

This way the length of the array doesn't have to be recalculated on
every iteration? Or am I completely wrong?

A clever compiler or hotspotter will likely do that optimisation for
you. Doing it yourself often just clutters the code. Optimise for
clarity, then if there is a problem, for speed.