After watching sites like Koders and Krugle grab some headlines for code search, Google has rolled its own code-specific engine, which is now available as a beta from Google Labs.
Code Search is the latest in Google’s efforts to embrace the developer community and give it useful tools. The search engine launched a repository for developer projects earlier this year. Now Google is lending its trusted brand to organizing a segment of online data that has traditionally been distributed across the Web.
The new search tool includes ZIP files as well as CVS repositories, and offers more precise special character searching. Unfortunately, the code search engine (still at the early beta stage) has a few problems, chief among which is that it doesn’t give the searcher any context for what it finds.
“It just searches indexed code from anywhere on the Web. Most of this is production code, not code snippets or tutorials,” said Chris McClelland, developer of AJAXPress. “The code you can get from code search could be wrong, a bad way to do things, simply not work, etc.”
His recommendation: Google needs to include code from tutorials and code snippet Web sites, such as this one, and tell searchers which sites are being indexed, perhaps even provide a quality score for each. Otherwise, searchers could accidentally copy bogus code indexed in the engine.
“Code snippet databases are much more useful,” he said, noting that much of the code returned is from open-source projects that are in continual development. “It’s like a textbook. [Developers] go for primary sources, not for what anybody will tell them.”
Google Code Search could also prove to be a tool for malware writers to distribute their code, he warned.
A representative from Google said Code Search was a labs project and that improvements were coming.
“Code that is contained in the body of a web page (such as a tutorial site) can generally be found in web search. Its reasonable to assert
we’d eventually want to include these sites in Code Search as well,” she said. Developers wishing to suggest improvements can contact Google at [email protected]
Commenters in the Google Group dedicated to Code Search said the tool looks very useful, but requested a few improvements. One commenter noted that ASP isn’t a language, and that the search should include Visual Basic and VBScript. Another commenter requested that ColdFusion be added to the index.
According to Google’s FAQ, Google goes to some lengths to determine what license(s) pertain to indexed code, but “understanding the legal requirements to reuse a piece of code is your responsibility.”