Have you ever wondered how your favourite programming language compares to others in terms of speed? How does for example Python compare to Ruby, PHP, Java or Perl? And how do all these scripting languages compare to actual compiled C++ code?

In most programming tasks, a central role is played by data structures such as arrays/lists, and hashes/dictionaries. I was curious how the performance of precisely such data structures differs across different languages. To test this, I wrote a series of basic data tests in all of the aforementioned programming languages, and I timed the execution of each test.

Besides a benchmark study, it was also a nice exercise to write essentially the same script in many different languages.

I will summarize some of the results of my experiments. I ran all tests 10 times on a Intel Core 2 Duo (Macbook Pro 5,3), running a 64-bit Ubuntu Linux 9.10. All times are in seconds and of course machine specific. C++ code was compiled with gcc 4.4.1, and for Java the Sun JDK 1.6.0 was used.

You can run all tests yourself, download the scripts and sources, inspecting the sources may also provide valuable insight in seeing how the tasks are implemented in the various languages.

Note that this is something I fairly quickly set up, so it may be that a few tests can be implemented more efficiently.

Test 1a/2a: Creating a new list of many integers:

Instantiating an array/list of two million integers (0 to 1999999). Note that all types except the C array are dynamically extendable.

Test 2c: Test if a value is in the list

Locating a value in the list, the value will be found half-way

Test 2e: Sorting a list

A sort is done on a list of two million random integers.

Test 2f: Polling all elements in a list

The value of each list item is retrieved during iteration over all items.

Test 2g: Setting all elements in a list

Set all items in a list to zero

Test 2h: Inserting 10 elements

Ten insertions in the middle of an array of two million.

Test 2i: Delete 10 elements

Ten deletions in the middle of an array of two million.

Test 3a: Creating map of two million integers (key==value)

Test 3b: Check if a key exists in map

Check if a key exists in the map: Less than 0.001 second for all languages

Test 3e/3f: Polling all elements in map

Retrieving the values of all elements

Test 3h: Setting all elements in map

Setting all elements in the map to zero

I made some small improvements in an existing gedit plugin for python code completion. The original post of the main author can be found here. My contribution consists of a few lines of code that enables the user to type unhindered by the window that pops up with suggestions. The window will automatically highlight methods and properties as you type along in the main window, and you can use the TAB key for autocompletion. My aim was to make the popup less obstructive and more transparent.

Download the modified code from GitHub.

I have launched a completely reworked version of my homepage. As you can see it features a fresh new layout and updated content. Another notable addition is a new multilingual video in my languages section.

The majority of the changes is technical in nature and can be found "under the hood" of the site, which is now written written in Python and Django, and on top of that using FeinCMS and django-diario.

I'm starting to more actively use FaceBook nowadays, FaceBook is quite a nice social networking site. I even wrote a small language application for the platform!

You can add me there, search for me under my real name, as that is the custom on FaceBook.