I have used computers to analyze recipes and track food intake for years now. Years ago I still used Microsoft Windows. There are some good proprietary programs in Windows that are useful for food analysis, but even the best aren't extremely well-documented. So once I discovered that the USDA food database is freely available, Microsoft Access seemed a natural way for me to develop my own application.
My Access application went through many iterations. Then I switched to Linux, which taught me one big lesson: never rely on expensive proprietary software. I poked around to see if there were any good Linux food analysis programs. The only well-developed one I could find, NUT, was a console program. At the time this scared me, so I just did not have a food analysis program for quite some time.
As I gained more experience with Linux, I discovered that there were Access-like applications for it, such as OpenOffice Base (which is still fairly new) and Kexi. But I did not trust these applications much: I generally found that huge office-suite programs for Linux are not very well documented and tend to be bloated. True, Access and Microsoft Office are bloated too, but at least there are tons of tutorials available on the Net. So I never touched OpenOffice Base or Kexi.
Though the Linux console scared me at first, as I used Linux more and more I realized how powerful and simple the console is. I came to appreciate its flexibility and even its beauty. So I took another look at NUT. Though I realized it is an excellent program, I couldn't quite get along with it. It ran in an console, but it was menu-driven and seemed to have arbitrary limits that I didn't like. There are Web-based programs for this purpose too, but I always find them cumbersome to use and poorly designed.
Then I slowly learned of command-line tools such as cut and join. I was surprised to learn that one can build a relational database management system using shell tools. I also was inspired by other command-line oriented tools that did things that I previously thought would be possible to do well only in a GUI. An excellent example of this is Ledger, a program that I now find indispensable for tracking my finances. So I decided to try writing a shell-based food analysis program; I called the program bashfood and, later, Shellfood.
Shellfood worked okay when it was small, but as it grew it became unmanageable. Shells are great for certain things, but building large programs is not one of those things. Eventually it became apparent to me that I would need to switch to a different language. Python seemed a natural choice, as it is often recommended for beginners. It is also a full-fledged language used by professionals--Linux always impresses me because tools that cost several hundreds of dollars for Windows are available free on Linux--better tools, too.
So, the result of several months of work is Pantry. It has evolved constantly as I have used and refined it. It works even better than I had envisioned, though others might find it hopelessly complicated. We'll see. Let me know what you think! Report bugs, too--I know they must be in here. Send emails to pantry-users@lists.sourceforge.net.
If you like Pantry, be sure to check out other command-line programs like Ledger and Todo.txt because they can make your life a lot easier. Good text editors like Vim and Emacs are indispensable, as is a nice shell like Zsh.