Having built the Mono version of nspectre recently, I thought about how I interact with the IDE (in may case the IDE I’ve spent most time in is Visual Studio.) Working on the Mono build of nspectre, I used MonoDevelop (with the Xml Editor plug-in) on Ubuntu Edgy and ran NAnt from the command line. When I’d built nspectre, I ran the example console program from the command line, too. To access CVS, I used LinCVS. So, I had a few windows open – with a decent size monitor, having a bunch of windows open is not the problem it would otherwise be. And it was easy to figure out what I was doing. It was comfortable. This realisation came as something of a shock to someone who has been used to doing nearly everything in the IDE.
I hear “integration with the IDE” as a frequent requirement for tools. What’s not to like about integration? I guess the problem is when the integration is such that it gets in the way of understanding what you are doing – or locks you into a particular or uncomfortable way of doing it.