I was introduced to Linux in 1992, when the first version was available. At the time I was visiting Poland and my company was building an inventory control system.

One of our partners had decided to use Linux to run their inventory system. What led them to use this little-known operating system for a critical task such as inventory control? Basically, they determined that it was their only viable option. DOS did not meet their needs for scalability and stability. Unix would have served their needs, but it was simply too expensive for them at the time. They had heard that Linux offered the scalability and stability of Unix, and figured it was worth their while to give it a try. It met their expectations, and they became very early Linux evangelists.

Upon hearing these partners praise Linux, I smirked. Based on my past experiences, I thought that all free software was poorly developed and unreliable. I certainly wouldn't trust it for something as critical as an inventory control system.

After I had completely discounted Linux, I noticed it slowly creeping into the development world. Developers at my software development companies in the U.S. and Poland brought in Linux machines here and there, and there was some buzz about Linux around the industry. Still, I didn't take it seriously. People were experimenting with these machines, but critical work was still performed on Unix and Windows systems.

Eventually, developers and system administrators started to take Linux seriously. I think that one key factor in this trend was Sun's 1991 shift from the SunOS (based on BSD) to Solaris. Many people who were previously happy with the SunOS -- myself included - were disappointed with Solaris' changes, and started to explore alternatives when it came time to upgrade or expand. Upon realizing that Linux was actually closer to the old SunOS than Solaris was, many SunOS devotees migrated to Linux.

By 1994, Linux was emerging as a serious development platform for C and C++. It seemed clear that Linux was a stable and viable development platform, and Parasoft was receiving numerous requests for a version of our Insure++ runtime error-detection tool that could help Linux developers find memory corruption, memory leaks, and other critical C/C++ errors. In response, we decided to port Insure++ to Linux, and released a version for Linux in 1995. It was well received by the Linux community, and it was the only tool of its kind on Linux for almost 5 years - a situation that is very rare in the software industry.

Based on the success of these initial efforts to produce development tools for the Linux community, we decided to port our other C and C++ products to Linux. We soon ported CodeWizard to Linux to help Linux developers automatically check whether their code complies with coding guidelines designed by gurus such as Scott Meyers. In 1999-2000, when we were developing the C++Test unit testing and coding standard analysis product, our Linux products were selling well and Linux was quite popular among C and C++ developers. Consequently, we designed this product to operate on Linux from the start. After Linux took the C/C++ world by storm, it started to catch on in the Java world. In response to this development, we ported Jtest, a Java unit testing and coding standard analysis product, to Linux in 2001.

Today, Linux remains a prime market for development tools. Linux developers are typically among the best developers; most take extreme pride in their code and are very receptive to tools that will help them produce better code faster. I think Linux will continue to be a good market for development tools unless the Balkanization of the market makes it impractical to produce development tools for all available flavors of Linux. With so many different Linux flavors being promoted by IBM, RedHat, Novell, and so on, development tool vendors either need to port applications to all of these versions of Linux, or settle with supporting only one or two. I hope that the industry reaches a resolution so that Linux development does not miss a beat.