|
|
php.net | support | documentation | report a bug | advanced search | search howto | statistics | random bug | login |
[2004-06-23 22:43 UTC] sprice at wisc dot edu
Description:
------------
I use include_once() and require_once() quite a bit, as
I would guess many others do. During a single page load,
I might call for the same file from many functions, just
to make sure things get defined correctly and in the
correct order. I did some testing today, and I noticed
that they both could be close to *two orders of
magnitude* faster for multiple inclusions.
I have attached the PHP code that I used to test this.
On my computer, when I use include_once() inside the for
loop, it takes about 6.0 seconds. When I use
improved_include_once(), it takes about 0.08 seconds.
The same goes for require_once().
Could/Should something like this be implemented in PHP?
It seems like a good idea, and you guys could do a more
efficient implementation than I am able to do.
Reproduce code:
---------------
function getMicroTime(){
list($usec, $sec) = explode( ' ', microtime() );
return ((float)$usec + (float)$sec);
}
function improved_include_once($filename){
if(!isset($GLOBALS['included_files'][$filename])){
include_once($filename);
$GLOBALS['included_files'][$filename] = TRUE;
}
}
$start_time = getMicroTime();
for($i = 0; $i < 10000; $i++){
include_once('my_file.inc');
}
echo (getMicroTime() - $start_time);
PatchesPull RequestsHistoryAllCommentsChangesGit/SVN commits
|
|||||||||||||||||||||||||||
Copyright © 2001-2024 The PHP GroupAll rights reserved. |
Last updated: Fri Dec 27 09:01:29 2024 UTC |
True, but that is why my function calls include_once() after its check. It checks for the most common case first, where people use it how I described previously. It was my hope that this function would only ignore files which include_once() would also ignore, and by doing this it takes care of the simple, common, cases. Anything more complex than this gets fully parsed by include_once(). I realize now that it is possible to change the underlying directory structure or change the current working directory in order to make include_once('my_file.inc') and include_once('my_file.inc') give different files. It seems a waste, though, that we can't speed up the function by several orders of magnitude because of a few edge cases. include_once() and require_once() are as slow as database accesses for me, even when the file has been included already. Maybe you could make it a two-level registry, and any filesystem writes or directory changes clears the first level. But then we may be getting a little complex. But it would be a nice performance increase for me (and others, I suspect).