Beautiful Code [153]
This system is now named Cryptonite and has been in constant off-and-on development and testing since then, in between other projects.
The first functioning prototype of Cryptonite was licensed to LFC as MailVault beta 2, and was open for testing in September 1999. It was the first OpenPGP-compatible webmail system available for public use and was almost immediately put to the test by LFC's investors and beta testers. Since that time, Cryptonite has evolved in many ways through interaction with users, the open source community, and the market. While not an open source product itself, it has led to the development of numerous components I decided to release as open source along the way.
11.1. The Heart of the Start
Developing Cryptonite and marketing and supporting associated services single-handedly for many years (with unwavering support and many invaluable ideas from my wife, Barkha) has been an incredibly interesting and rewarding journey, not only from a development perspective but also from an entrepreneurial one.
Before jumping into the nitty-gritty of the system, I thought I'd touch upon some points that have impressed themselves strongly in my consciousness over the course of this project:
My friend Rishab Ghosh once quipped that there's a lot of hype about how the Internet can enable wired hackers to work from anywhere, but most of the people who create this hype live within a small area in California. The great thing about an independent startup project is that it really can be done anywhere, and dropped and picked up again when convenient. I've hacked on Cryptonite over many years on four continents, and it may well be the first high-quality software application developed in large part in the Himalayan mountains. (I used the word "wired" before loosely. In reality, five wireless technologies facilitated our connectivity in the Himalayas: a VSAT satellite Internet link, Wi-Fi, Bluetooth, GPRS, and CDMA.)
When working on a project as a single developer in your spare time, remember the old hacker wisdom that "six months in the lab can save you ten minutes in the library." It's critical to maximize your reuse of existing code libraries. For this reason, I elected to develop the system in Perl, a popular and flexible high-level language with a rich library of mature, free software modules, and the Perl hacker's first virtue of laziness informed every design decision.
Especially for end-user application software, ease of use is a critical issue. It is essential to the function of such code to present a simple, accessible interface to the user. The usability considerations of developing an end-user security application are even more significant, and were in fact a key factor in making the Cryptonite system worth developing.
To get off to a running start, it's a good idea to implement a working prototype first, and use a prototype-to-production path to move to production deployment after the essential functionality is implemented. This can be a huge help in getting the basic design and structure right before you unleash the code on hundreds or (hopefully!) millions of users.
Keeping your system as simple as possible is always a great idea. Resist the urge to get suckered into using the latest complex buzzword technology, unless the application really demands it.
Processors are pretty fast now, and programmer time is generally more valuable than processor time, but speed is still critical for application software. Users expect their applications to be snappy. For web applications, which many users will use concurrently, investing some time in optimizing for speed is a Very Good Thing.
A software application is a living entity, in constant need of attention, updating, enhancement, testing, fixing, tweaking, marketing, and support. Its success and beauty in an economic sense depends directly