I got a close look at Krugle (rhymes with Google), the forthcoming (March 8) vertical search engine for programmers. The search engine finds content from a constrained set of domains, starting with the main Web sites that programmers use and rippling out to sites that are associated by strong interlinking. Krugle also analyzes site content linguistically--based on about 500 technology words--determining the relevance of the content based on a ratio of the keywords to all the words on the page or site, according to Steve Larsen, CEO and co-founder of the company.
In addition, Krugle searches for chunks of code on pages. The company is also indexing code from a variety of sources, including from Sun, O’Reilly, IEEE and the ACM. For sites that require registration, Krugle is working with those companies to integrate with their registration systems. If the user is not a registered Sun developer, for example, and clicks on some Sun data behind the firewall, Sun’s developer log-in/new user screen will pop up. The source code rendered by the search engine is parsed and retains the code structure and formatting appropriate to the language. The metadata of code projects, such as licenses, operating systems and size of the bug database, is also available as well as API reports on code. You can also view where a project exists within the universe of code.
Steve Larsen, CEO, and John Mitchell, chief architect, of Krugle
Krugle also allows registered users to comment and annotate code in layer that floats above the search results. Users can also tag and download projects as well as bookmark lines, pages and projects.
The search engine will be free and ad supported. A paid version will be available as well that allows developers to plug Krugle into IDEs like Eclipse and index code and technical information behind corporate firewalls. Larsen told me that Krugle will have an enterprise version next year that will allow companies to index to all their code using Krugle's platform.