On 18/02/11 17:35, bytespider wrote:
> for (var i = 0, length = arr.length; i < length; ++i) {
...
> Note I also use ++i over i++ because
> it's slightly faster on some modern browsers but mostly out of habit.
Okay, now we're firmly in micro-optimization land, but I'm curious...
partially because this topic comes up once in a while, and partly
because I also tend to use ++i rather than i++ (purely out of habit).
This used to be a sensible optimization in C once, a long time ago, but
modern compilers like gcc or javac will automatically optimize an
expression like this when its return value is not used (I've appended a
few examples in case you want to confirm it for yourself). As far as I
know, there is no longer any point in preferring pre-increment over
post-increment in a for-loop like this. Same thing goes for C++, as long
as i is a simple type.
Now you say that modern browsers will perform _better_ when you use the
pre-incement. Did you actually test this? I tried to verify it with a
couple of modern engines, but didn't see any difference. I didn't expect
to see any - modern JS engines should have more advanced optimizers than
older ones; they should be better able to recognize that the return
value won't be used and the expression i++ can be optimized. It seems
like most modern engines do this.
>From a semantic viewpoint, some people say that ++i expresses the intent
more accurately. From a logical viewpoint, using ++i should never be
slower than i++, regardless of optimizers. From a pragmatic viewpoint -
who cares, they're both fast enough. As I said, I'm just curious.
Before I get flamed: this is a purely academic question. I'm certainly
not suggesting it has any impact on real life JS performance from our
perspective.
stefan
_____
$ cat pre.c
#include <stdio.h>
int main ()
{
int i;
for (i = 0; i < 5; ++i) {
printf("Line %d\n", i);
}
return 0;
}
$ cat post.c
#include <stdio.h>
int main ()
{
int i;
for (i = 0; i < 5; i++) {
printf("Line %d\n", i);
}
return 0;
}
$ cat pre.java
class Test
{
static public void main (String args[])
{
int i;
for (i = 0; i < 5; ++i) {
System.out.printf("Line %d\n", i);
}
}
}
$ cat post.java
class Test
{
static public void main (String args[])
{
int i;
for (i = 0; i < 5; i++) {
System.out.printf("Line %d\n", i);
}
}
}
$ gcc < pre.c -xc -S -o pre.s -
$ md5sum pre.s
d9b04afd3a0c5e95d3492459f736e7ec pre.s
$ gcc < post.c -xc -S -o post.s -
$ md5sum post.s
d9b04afd3a0c5e95d3492459f736e7ec post.s
$ cp pre.java Test.java
$ javac Test.java
$ md5sum Test.class
c787c1063aa3d4164d9f2b63b7df8fd2 Test.class
$ cp -f post.java Test.java
$ javac Test.java
$ md5sum Test.class
c787c1063aa3d4164d9f2b63b7df8fd2 Test.class
--
LOAD"Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn!",8,1
RUN!
--
To view archived discussions from the original JSMentors Mailman list:
http://www.mail-archive.com/jsmentors@jsmentors.com/
To search via a non-Google archive, visit here:
http://www.mail-archive.com/jsmentors@googlegroups.com/
To unsubscribe from this group, send email to
jsmentors+unsubscr...@googlegroups.com
