Windows 2000 code size

Here's what John Dvorak said about Windows 2000 in PC Magazine on April 6, 1999:

"With a base of 50 to 60 million lines of code that obviously nobody has a grip on, there has to be an intense feeling of panic within the company [Microsoft]. Silverberg, who brought out Windows 95, is seen as a savior, but he must know the reality of the situation too. I've talked to many ex-Microsoft folk who all tell me that nobody has a handle on Window's code. It's completely out of control - a hodgepodge of objects and subsystems nobody fully understands."

What John is referring to is the fact that when code becomes this large nobody has a clear idea of how it works as a whole. Individual programmers can only know a very small portion of the whole. It takes hundreds or thousands of people to maintain the code but the whole of the code cannot be managed. One reason is that it's difficult to manage that number of people. This is a house of cards that will come tumbling down eventually.

In striking contrast, the Linux code base is estimated to be about 500,000 lines of code. This is less than one percent the size of Windows 2000!
We don't include the Linux GUI (X Window) as part of the base code because with Linux the GUI is optional. The GUI adds another 500,000 lines of code. Linux and X Window together are about 1.8 percent the size of the Windows 2000 code base.

Here's what John Dvorak said about Linux in the same article: "The way I see it, Linux's code base is under much tighter control than Windows."

Back