I do find it interesting that Andy Hunt Ruby and Rails so much considering that Rails is one of the most magical and mysterious code generators I’ve ever encountered in my life. Rails is great if you’re building a certain kind of web application but there’s so much mystery behind the scenes it can be unsettling, especially for a newcomer to the framework.

And I’ll defend knowing one text editor well. I write some code in an IDE, but a solid text editor lets me tear through massive amounts of data, peak at code files, data transformations, analysis, quickly write scripts, or to take notes. An IDE is only good for writing code the IDE supports (and jack of all trades IDEs like Eclipse are just awful). Plus a good text editor is going to have syntax highlighting, snippets, macros, regex support, and lots of other indispensable power tools. The IDE is great, but proficiency with a solid text editor is still essential.

Maybe it really depends on the language (I’m thinking here of LESS, which generates CSS), but it seems like today’s major languages already have enough built-in mechanisms that you don’t need to go and create some program to duplicate your code at compile time.

I have a different viewpoint about generated codes though. Like all powerful things in the world, it could granted you infinite power, or infinite misery if you are not careful. I think the specific case that generated code bite us hard is when each developer generate his own code and checks in to SVN. And then you have to go through all the pain of merging them. I think what the two refers to was to use a build server to do the generation (this be the only official version of the generated code), nobody checks in the generated source and use the generated code as if it is a library. I think the best documentation for the generator would be its test methods (if you do TDD, that is).