I noticed the other day that the pages on this site had double body tags in the HTML. As someone who favors standards it was kind of embarrassing. But what started as a quick fix turned into a few days of ruminating about how I’m putting this site together. Read the rest of this entry »

Having worked for Folio and NextPage, companies that have specialized in full-text indexing using custom search engines, I have an admitted knowledge liability with regard to relational databases. It never really made sense to have a relational database and our own search database coexisting in an application. But it never occurred to me that such a knowledge liability could be considered a benefit in terms of thinking about how applications should be designed. Read the rest of this entry »

This is a continuation of thoughts after reading Design by Contract by Example by Richard Mitchell and Jim McKim. This time I want to write a little about design by contract without direct language support. What do you give up without language support, and is it still worth it to struggle through defining contracts without that support? Read the rest of this entry »

I recently finished reading Design by Contract by Example by Richard Mitchell and Jim McKim. The book, as its title suggests, gives advice about writing good contracts through a set of principles applied in various examples. In addition to wanting to learn more about design by contract generally, I came to this book with a couple questions: Can you do design by contract effectively without built-in language support, and what is the relationship between design by contract and unit testing? In this article I’ll talk about design by contract generally, and discuss those two questions in follow-on articles. Read the rest of this entry »

Updated September 8, 2005

Logging is a testing and diagnostic technique that lies somewhere between automated testing and debugging. It can help you understand what your application is doing, especially in cases where there is some non-determinism because of threading, or when running in an environment where debuggers are not an option (such as on a customer’s machine). This article is an attempt at defining good logging style. There are lots of things already written about the technical aspects of using a particular logger, like Log4J, but not so much about how to use logging effectively. Read the rest of this entry »

How do you get your production code to use real objects and your test code to use mock objects? Read the rest of this entry »

I’m occasionally surprised that some programmers prefer an error handling system of returned error codes over exceptions. After spending several years with both approaches, I’ve become convinced that exceptions are a superior model. If you agree, great! If not, read on and let me try to persuade you. I’m always up for a discussion on the subject, too. Read the rest of this entry »