>>I'm asking this on behalf of a friend who is considering starting a large >>transcription project, entering music into abc. He wants to use Gore >>or Breathnach's melodic codes to simplify searching. The question is, >>what field to use? >> > > > The field following the K: field, of course. I'm not being >sarcastic. The information is in the abc already; it's just a question of >how to digest it for fast searching. (In a sense, an abc tune is already >organized as a database.) I don't know Gore's system, but Breathnach's >involved simplifying the first measure or so of the tune, and was >intended for a search-by-hand. This should be easy to automate: one >should be able to write something to generate this code from the abc, and >I suspect that with the present processor speed, you'd have to search a >pretty large file--or the whole web--before it'd be a lot faster to put >this into the headers than to simply generate it on the fly. I'd suggest >putting some serious thought in how to generate this code efficiently. >But I'd also suggest looking into Chris Walshaw's scheme (see the ABC >indexing guide, pp 3--5) which was designed to facilitate searching of >abc files. It works pretty well, and might have its own advantages.
You're right, of course. Actually the whole point of these melodic codes is that you don't attach them to the data itself, you keep them in a separate sorted list so that you can do a binary search on them. I wasn't really thinking last night:-) Phil Taylor To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
