Yesterday afternoon, I was lurking about the exhibition hall and stood in a bit for the code search demonstration that OReilly was giving. I immediately asked why they didn’t just use krugle, and they had a pretty good answer – they wanted a walled-garden search of the books – not just all open source projects. The theory presented was that the code examples in the books had some level of expectation of being better than your joe-random open source project.
They did a neat sort of concordance (only a real book geek could love that, even if it is useful…) in your standard “larger type == more popular” layout that has been so popular for tag sets.
Their code search was sort of brutally slow, and I didn’t feel like it had a very good index (the search term “python json” didn’t return anything), but the interface that was wrapped around it was very intriguing. They’ve got a lot of interface tweaks and friendliness wrapped in there that’s not obvious at first breath.
I rather hope we see the optimized version of this hitting the Safari book library before too long…