[dmunoz]

I'm interested in this as well. Working on tooling to make this a more efficient process is something that I daydream about often, so it would be interesting to hear some perspectives.

Do you generate tags and hop around functions as you discover them? Do you do a high level overview of the important bits first, and then dive in later? I am sure it is different for everyone, but would still be interested to read about others approaches.

I've had this idea for awhile about bringing up a project in a methodical way in its source control history, and adding new features in a similar manner, so that it would be easier to introduce a codebase to those starting out in it. The problem with lots of large codebases is that as time goes on, configuration details or special cases are added on that obscure the meat of the program. If it were trivial to see the early version, and its evolution, it would be easy to pick out the mental model of what the code actually does.