Some government-funded open source offers an alternative bug-fixing approach to that used by Coverity, whom we wrote about yesterday.

The NIST, working with researchers at UT-Arlington, plan to release open source code for what they call automated “combinatorial” testing.

While Coverity’s software tests lines of code for common mistakes, like failing to close quotes or basic processes, the new code tests a variety of error states in combination. (A release on this is at Science Daily.)

By testing the interaction of six variables simultaneously, the authors say, they can catch most error states and fix most bugs as the program is working, before it goes into beta test.

The NIST and Coverity techniques can be seen as complimentary, the former testing for problems as the code runs, the latter testing for problems in the code itself.

The technical term for this is Automated Combinatorial Testing for Software (ACTS) and a collection of papers on it is available through the NIST. A presentation on the concept is also available.

While it’s possible proprietary vendors will adopt these and other government-funded testing breakthroughs, at some point I think you have a critical mass of open source tools to give those vendors a run for their money. Especially for smaller shops and those with smaller budgets.

Not that there’s anything wrong with that.