I've been using the UNIX dinosaur RCS to manage version control of my text files. Everyone I've talked to thinks that this is crazy, given that there are fancier, more comprehensive solutions out there, especially Subversion.

But out of some strange obstinacy, I've stuck with the simpler RCS for managing my notes, papers, etc. And I just discovered that this crusty old program works really well for managing my academic work flow.

Let me explain:

One of the legendary disadvantages of RCS is that is works on a file by file basis. Version control is completely decentralized. When you check in a file, RCS creates a version control file with the same name. (If you create a RCS directory in the directory you're working in, it deposits the version control file there). This is terrible for much development and collaborative, because there's no easy way to rename files. Moreover, the different files all have individual version numbers, so there's no versioning of an entire directory.

But suddenly I realized that this disadvantage is a huge advantage for my writing workflow.

1) I begin a paper as an outline using VIM outliner and the otl extension. So the file gets checked into version control as an outline and I can keep track of my changes to the outline.

2) When I'm ready to work the outline into a draft, I change the extension on the file to txt. Thus, when I check this file in with RCS, it creates a new version control file with the txt extension, leaving the former otl version control in a separate self-contained state. I hack away at the outline with vim, and all changes get monitored by the new version control file.

3) Then, when I'm ready to publish the draft and shift it over to LaTeX for formatting, I change the extension to tex. Once again, when I check it in, I get a new version control file with the tex extension that keeps track of all of my markup changes. Usually, all further revisions of the paper are done with the LaTeX markup present.

The result: three version control files that represent distinct stages in the history of my paper. Outlining (paper.otl,v). Drafting (paper.txt,v). Polishing (paper.tex,v).

One further advantage: because there's no central repository with RCS, these three "histories" are not only portable, but also easily searchable using grep.

So the old dinosaur has some life in it after all...