I guess from a performance perspective a simple string comparison and replace might be more effective.

Bob repies: I use regex all the time, but tend to avoid regex when a simple comparison will do. Not only for performance reasons, but for clarity. But the main motivation for this post was running across this logic for the umpteenth time. I know the intent, but it’s one of those sneaky things that, depending on the data being fed to it, is apt to work most of the time and then occasionally fail. That is the worst class of bugs.

Bob responds: Yes, I think the suggested pronunciation of “ScamL” indicates that Petzold is poking fun at a someone’s efforts in this space. Let’s hope it’s doesn’t end up hitting too close to home.

I am glad that they have error codes, severity, state and the like, because that can be used to gain additional information (i.e. easier to search for on google). But diskspace is cheap, and a couple dozen extra characters in a log entry is not going to fill it that much faster.

I can’t decide if it is laziness or incompetence, but this seems to exemplify their attitude toward error reporting. (Although it’s not as bad as “Catastrophic Failure” in Visual Studio).