23 Aug 2000 Pseudonym   » (Journeyer)

A few diary entries ago, Skud asked what was so "extreme" about XP. This is a partial answer to that question. I'm putting it here and making it more general because others might be interested.

This was useful for me because I've been interested in software development theory for a long time now, despite not having formally studied software engineering apart from one ill-fated semester which constituted my lowest mark in my computer science degree. Putting things into words makes you see just how much you understand (or don't understand, as the case may be) something.

Extreme Programming is a bit like design patterns, refactoring, programming on purpose, or Unix for that matter. It seems so much like common sense. However, there are two realities at work here.

The first is Henry Spencer's principle:

Those who do not understand Unix are condemned to reinvent it, poorly.

The principle extends to a lot of things beside Unix. In general, good practice requires good theory. If we don't have a good reason to do something, we might as well be doing something else instead. Software engineering is no different. From the point of view of the developer, IMO the most important legacy of The Cathedral and the Bazaar is that it tried to explain for the first time why bazaar development works. Once we know that, we can decide what bits are important, what bits aren't important and what experimentation might be fruitful. XP is no different. It may look like common sense, but in reality, it ain't so common. We need to understand it so that we aren't condemned to reinvent common sense, poorly.

The second reality is that medium- and large-scale software development in the Real World(tm) of deadlines and budgets requires justification to those who hold the schedules and the purse strings. Giving common sense a name and a series of papers and books makes it easier to justify to the PHBs. This goes doubly for the quality accreditation world (e.g. ISO 9000 and its relatives), where it's not so important what big-M Methodology you use, but the fact that that you use one at all. It puts you that one crucial step above ad hoc in the eyes of the professional world.

So why is it Extreme? It's extreme because it carries principles which are well-known and considered common-sense in the formal software development world and takes them to extremes. XP literature often admits that they may appear almost too simple. Well, they are simple. They are, however, implemented to extremes.

Let's look at a few examples.

  • We all know that design is a good thing. We've been taught that good design includes doing a big design up front so that the code falls out naturally. XP teaches that design is so good that we should do it all the time. Only when you have working code can you truly tell how it should be internally structured. The principle is called Refactor Mercilessly.
  • We all know that code reviews are a good thing. We've been taught that you should get someone else to look over your code before it gets checked in. XP teaches that code reviews are so good that we should do it all the time. In fact, programmers should work together. This principle is called Pair Programming.
  • We know that talking to the customer is a good thing. XP teaches that it's so good that we should get a customer to sit with the development team throughout the course of the project. This principle is called Onsite Customer.
  • We know that adding people to a late project makes it later. This is one of Brooks' laws from The Mythical Man Month. XP teaches that adding time to a programmer's week makes the project later too. XP defines OverTime as the time spent working that makes a programmer less productive in the long term. OverTime is banned in XP.

We in the open source community know a lot of this already, even if we can't articulate it. XP comprises a lot of bazaar principles such as Collective Code Ownership, Continuous Integration and Frequent Releases. But there are other principles, such as Pair Programming and Onsite Customer, which rely on proximity of developers and customers. While bazaar developers probably can't make use of these, open source companies should read and learn.

As always, this is IMHO and YMMV, but I hope this helps answer the question.

Latest blog entries     Older blog entries

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!