Viewed   70 times

I'd like to find a way to determine how long each function in PHP, and each file in PHP is taking to run. I've got an old legacy PHP application that I'm trying to find the "rough spots" in and so I'd like to locate which routines and pages are taking a very long time to load, objectively.

Are there any pre-made tools that allow for this, or am I stuck using microtime, and building my own profiling framework?

 Answers

1

I have actually done some optimisation work last week. XDebug is indeed the way to go.

Just enable it as an extension (for some reason it wouldn't work with ze_extension on my windows machine) , setup your php.ini with xdebug.profiler_enable_trigger=On and call your normal urls with XDEBUG_PROFILE=1 as either a get or a post variable to profile that very request. There's nothing easier!

Also, i can really reccommend webgrind , a webbased (php) google Summer Of Code project that can read and parse your debug output files!

Thursday, November 17, 2022
1

I agree with the comment, what you are doing is very underhanded, but after 10 years in this biz I can attest to one thing: Half the code you get is so convoluted it might as well have been minified, and really function/var names are so often completely arbitrary, i've edited minified js and it wasn't much more of a hassle than some unminified code.

I couldn't find any such script/program, most likely because this is kind of against the PHP spirit and a bit underhanded, never the less.

First: Php isn't white space sensitive, so step one is to remove all newlines and whitespace outside of string.

That would make it difficult to mess with for the average tinkerer, an intermediate programmer would just find and replace all ;{} with $1n or something to that effect.

The next step would be to get_defined_functions and save that array (The 'user' key in the returned array), you'll need to include all of the files to do this.

If it's oo code, you'll need get_defined_classes as well. Save that array.

Essentially, you need to get the variables, methods, and class instances, you'll have to instantiate the class and get_object_vars on it, and you can poke around and see that you can get alot of other info, like Constants and class vars etc.

Then you take those lists, loop through them, create a unique name for each thing, and then preg_replace, or str_replace that in all of the files.

Make sure you do this on a test copy, and see what errors you get.

Though, just to be clear, there is a special place in hell reserved for people who obfuscate for obfuscation's sake.

Check out: get_defined_functions get_declared_classes and just follow the links around to see what you can do.

Saturday, November 26, 2022
1

Hey, I use XDebug and KCachegrind to check out whats going on server side.

http://kcachegrind.sourceforge.net/html/Home.html

http://www.xdebug.org/

They are pretty easy to set up and have been invaluable to me in identifying bottlenecks in my code.

Monday, November 28, 2022
 
merv
 
4

Under Windows it seems that files in vendor/bin are actually batch files, invoking the original file (and not php files which phpdbg will understand).

In this case:

dir=$(d=${0%[/\]*}; cd "$d"; cd "../phpunit/phpunit" && pwd)
                                  ^^^^^^^^^^^^^^^^^^
"${dir}/phpunit" "$@"
 ^^^^^^^^^^^^^^

i.e. ../phpunit/phpunit/phpunit (this path is relative to vendor/bin); thus the actual file is at vendor/phpunit/phpunit/phpunit.

And you can invoke it directly via phpdbg -qrr vendor/phpunit/phpunit/phpunit then.

Friday, December 23, 2022
5

After some more exploration into performance profiling, I have discovered that using a Stopwatch is not an accurate way to measure the performance of a particular task

(Thanks hatchet and Loren for your comments on this!)

Reasons a stopwatch are not accurate:

  1. Measurements are calculated in elapsed time in milliseconds, not CPU time.
  2. Measurements can be influenced by background "noise" and thread intensive processes.
  3. Measurements do not take into account JIT compilation and overhead.

That being said, using a stopwatch is OK for casual exploration of performance. With that in mind, I have improved my profiling algorithm somewhat.

Where before it simply executed the expression that was passed to it, it now has the facility to iterate over the expression several times, building an average execution time. The first run can be omitted since this is where JIT kicks in, and some major overhead may occur. Understandably, this will never be as sophisticated as using a professional profiling tool like Redgate's ANTS profiler, but it's OK for simpler tasks!

Wednesday, September 14, 2022
 
cvng
 
Only authorized users can answer the search term. Please sign in first, or register a free account.
Not the answer you're looking for? Browse other questions tagged :