Performance advantages of using methods inside of classes verses data structures with libraries of functions?

Question

Basically is the only advantage of object oriented languages the improved understanding of a programs purpose?

Do the compilers of object oriented languages break apart the objects into structures and function libraries?

Answer 1

Yes, method code is central to the class definition and each instance method accepts an implicit this pointer to the data as its first argument. If you disassemble an instance method call you will see this.

Answer 2

Basically, yes. The only advantage is improved understanding of code.

For some languages the OO version is the same as the non-OO version after compilation. Perl for example. For the majority of cases the OO version is much slower than the non-OO version. With very rare exceptions, non-OO languages are always faster than OO languages.

But in general, most experienced programmers will tell you not to worry about the performance differences between OO and non-OO languages (or Lispers will tell you not to worry about the performance difference between procedural and functional languages). This is because you should never, ever, ever, underestimate the importance of understanding code.

These days we rarely talk about it anymore because we've gotten used to using very high level languages - be it OO or functional or multi-paradigm or metaprogramming. But back in the 80s and 90s there was what was then known as the software crisis. What was the software crisis? It's basically the fact that most software projects were never completed!

The software crisis affected all sectors of the industry: from military radar systems, to games to commercial operating systems. The consumers called them vaporware. They were projects that were too ambitious.

But these days there are lots of very ambitious and impressive projects that manage to reach at least beta versions (and for web2.0 beta is good enough for public consumption). Part of the reason is that we now understand requirements engineering better and we also understand the process of software development better. But part of it is also because we have better tools to actually understand what we're doing. And OO is part of that toolset.

Answer 3

Here are a couple of links to compare speed, first is comparing C/C++, please read the entire article: http://unthought.net/c++/c_vs_c++.html

To compare Python, Java, C++, PHP and other languages:

http://blog.dhananjaynene.com/2008/07/performance-comparison-c-java-python-ruby-jython-jruby-groovy/

But, to answer your question, the main advantage of OO is that for many problems it is the best way to model the solution, as the model naturally fits into objects. But, if you try to force it to work where it is not a good fit you will have harder to understand code.

There are various language paradigms as there are many different types of problems, and you should pick the language type that best models the solution. For example, I would not want to write an OS in C++ as it doesn't seem to fit well in OO methodologies, but I would also not want to write a car racing game in C, as it would make more sense to have objects.

Depending on the language and compiler, you may see the compiled application compile to C, but others are not, as some are going to be interpreted.

For example, C++ compiles to C, but Java does not, neither does .NET languages. PHP is generally interpreted, though it is possible to compile it (though I have never tried it). One compiler is:

http://www.phpcompiler.org/