Developer's Toolkit

Blog archive

Good Code or Code That Works

At VSLive! Boston last week, I also sat in on a session by Jackie Goldstein on worst practices in coding. What he did was to display code examples and ask the audience what was wrong with them. The code worked, but was slow or inefficient or both, and Jackie led the audience on a journey of discovery of poor coding.

The interesting thing is that most of his examples depended on having an understanding on how underlying mechanisms worked – usually the .NET Framework. He also illustrated differences between the .NET Framework 1.1 and 2.0 that has an effect on coding practices when moving from one to the other. The common thread was that writing good code required knowledge of more than just the programming language (C# or VB.NET, in this case).

Joel Spolsky ( has pointed out on multiple occasions that writing code requires a significant understanding of all layers of abstraction, from the programming language down to the hardware. He has lamented the fact that it is possible to graduate with a degree in computer science from an otherwise reasonable school by just using Java, and not understanding the concept of pointers (

I like abstractions. They tend to simplify otherwise difficult concepts, and accelerate productivity for programmers. But I found myself nodding in agreement with both Jackie and Joel, in that knowing your programming language is not enough. I have never encountered the situation that Joel describes (, in which programmers simply do not get pointers, but I can see it happening without a proper grounding in languages.

You might make the argument that such an understanding is nice to have, but no longer a necessity, because languages have moved beyond the need to directly manipulate the data in specific memory locations. There is a certain amount of truth to that. Millions of Java and VB programmers are a testament to the ability to write software effectively without pointers. And Jackie's .NET examples certainly did not require the use of pointers.

But the point that you need a deep understanding of everything the computer is doing still holds. Jackie's poor code did run, and was correct, but use managed memory poorly, or used an inappropriate .NET class or method. The fact remains that if you do not know what the underlying layers are doing, you will make the wrong programming decisions.

Posted by Peter Varhol on 10/29/2006 at 1:15 PM

comments powered by Disqus


  • Uno Platform Ports Windows Calculator to Linux

    Uno Platform has ported the famed Windows Calculator, open sourced last year, to Linux as part of a continuing "proof point" effort to demonstrate the reach of what it describes as the sole UI offering available to target Windows, WebAssembly, iOS, macOS, Android and Linux with single-codebase applications coded in C# and XAML.

  • ASP.NET Core OData 8 Preview Supports .NET 5, but with Breaking Changes

    ASP.NET Core OData, which debuted in July 2018, is out in a v8.0 preview that for the first time supports the upcoming .NET 5 milestone release.

  • VS Code Java Team Details 5 Best Dev Practices

    Microsoft's Visual Studio Code team for Java development added a new Coding Pack for Java installer and detailed best practices for setting up a development environment.

  • Binary Classification Using PyTorch: Defining a Network

    Dr. James McCaffrey of Microsoft Research tackles how to define a network in the second of a series of four articles that present a complete end-to-end production-quality example of binary classification using a PyTorch neural network, including a full Python code sample and data files.

  • Blazor Debugging Boosted in .NET 5 RC 2

    In highlighting updates to ASP.NET Core in the just-launched second and final Release Candidate of .NET 5, Microsoft pointed out better debugging for Blazor, the red-hot project that allows for C# coding of web projects.

Upcoming Events